From eng-partner-management at redhat.com Tue Oct 1 16:29:16 2013 From: eng-partner-management at redhat.com (Engineering Partner Management) Date: Tue, 01 Oct 2013 12:29:16 -0400 Subject: [rhelv6-list] Red Hat Launches the Next Release of Red Hat Enterprise Linux 5 Message-ID: <524AF85C.3040707@redhat.com> Red Hat Enterprise Linux 5.10 improves stability, security for enterprise applications Raleigh, N.C. ? October 1, 2013 ? Red Hat, Inc, (NYSE: RHT), the world's leading provider of open source solutions, announced today the release of Red Hat Enterprise Linux 5.10, the latest minor release of the mature Red Hat Enterprise Linux 5 Platform. With an emphasis on providing greater stability for critical applications, Red Hat Enterprise Linux 5.10 offers enhanced features for reliability and security, including an updated version of OpenSCAP ? the open source Security Content Automation Protocol (SCAP) configuration scanner, which meets the National Institute of Standards and Technology?s (NIST) SCAP 1.2 standard. Red Hat Enterprise Linux 5.10 reiterates Red Hat?s commitment to a 10-year lifecycle for Red Hat Enterprise Linux 5 and maintains backwards compatibility with all previous releases of Red Hat Enterprise Linux 5, including supported hardware and software platforms. Beyond OpenSCAP, Red Hat Enterprise Linux 5.10 also includes: * MySQL 5.5, the most recent, stable version of that open source database. MySQL 5.5 includes a number of improvements in terms of speed, scalability, and ease of use. For customers' convenience, Red Hat Enterprise Linux 5.10 also includes MySQL 5.1, which is required in order to upgrade to MySQL 5.5. * Enhanced Subscription Management Tools, which now provide more insight into how customers can optimize and benefit from their Red Hat subscriptions. With Red Hat Enterprise Linux 5.10, customers can better match subscriptions to their unique system needs, better report on and manage subscription usage across a large inventory of systems, and add custom, searchable key value pairs to systems, all without having to go through the Red Hat Customer Portal. * Red Hat Access, delivered through the Red Hat Support Tool, provides an integrated, seamless way to get answers with exclusive Red Hat Knowledge, use Red Hat?s automated diagnostic services for problem determination and engage with Red Hat directly from a Linux terminal. Customers can use Red Hat Access to quickly resolve issues and analyze log files, core dumps and error messages, all from within Red Hat Enterprise Linux 5.10. Additionally, Red Hat Enterprise Linux 5.10 customers with Red Hat Enterprise Linux Developer subscriptions will now gain access to Red Hat Developer Toolset 2.0 (http://www.redhat.com/about/news/press-archive/2013/9/red-hat-releases-red-hat-developer-toolset-2-0-with-update-to-gcc), which offers users a selection of popular, open source developer tools. Red Hat Developer Toolset 2.0 has a more frequent release cadence and exists on a separate lifecycle from Red Hat Enterprise Linux, enabling developers to take advantage of the latest stable open development tools for application innovation. Additional Resources * Download Red Hat Enterprise Linux 5.10 https://www.redhat.com/wapps/sso/login.html?redirect=https://rhn.redhat.com/rhn/software/channels/All.do * Learn more about Red Hat Enterprise Linux http://www.redhat.com/products/enterprise-linux/ * Learn more about the Red Hat Enterprise Linux lifecycle https://access.redhat.com/support/policy/updates/errata/ Connect with Red Hat * Learn more about Red Hat http://red.ht/IOS5vm * Get more Red Hat news http://bit.ly/I6BrYW * Subscribe to the Red Hat news RSS feed http://red.ht/JVLT3Y * Follow Red Hat on Twitter http://bit.ly/JnIsUg * Join Red Hat on Facebook http://on.fb.me/JVGXMi * Watch Red Hat videos on YouTube http://bit.ly/JEkzvc * Join Red Hat on Google+ https://plus.google.com/+RedHat About Red Hat, Inc. Red Hat is the world's leading provider of open source software solutions, using a community-powered approach to reliable and high-performing cloud, Linux, middleware, storage and virtualization technologies. Red Hat also offers award-winning support, training, and consulting services. As the connective hub in a global network of enterprises, partners, and open source communities, Red Hat helps create relevant, innovative technologies that liberate resources for growth and prepare customers for the future of IT. Learn more at http://www.redhat.com. From sjoe302 at yahoo.co.uk Sat Oct 5 08:58:46 2013 From: sjoe302 at yahoo.co.uk (Joe Smith) Date: Sat, 5 Oct 2013 09:58:46 +0100 (BST) Subject: [rhelv6-list] rh6 sys admin for cambridge uk Message-ID: <1380963526.85562.BPMail_high_carrier@web172105.mail.ir2.yahoo.com> Large scientific installation. http://www.embl.de/aboutus/jobs/searchjobs/index.php?newlang=1&ref=EBI_00329&back=%2Faboutus%2Fjobs%2Fsearchjobs%2Findex.php%3Floc%3D2%26list%3D1 International atmosphere. From eng-partner-management at redhat.com Tue Oct 8 18:31:02 2013 From: eng-partner-management at redhat.com (Engineering Partner Management) Date: Tue, 08 Oct 2013 14:31:02 -0400 Subject: [rhelv6-list] Latest Beta Release of Red Hat Enterprise Linux 6 Now Available Message-ID: <52544F66.7090304@redhat.com> Today, we are pleased to announce the beta release of Red Hat Enterprise Linux 6.5. To keep pace with the demand for enterprise-class infrastructure, the newest version of the leading Linux platform extends its scalability and manageability to aid in the build-out and control of large, complex IT environments. For example, Red Hat Enterprise Linux 6.5 will be designed to simplify the operation of mission-critical SAP applications by automating the optimal configuration of common SAP deployments. The beta release of Red Hat Enterprise Linux 6.5 also offers enhancements and new capabilities in the following key areas: Subscription Management Services -------------------------------- Red Hat Enterprise Linux seamlessly integrates with Red Hat subscription management services to manage the local system and its allocated subscriptions. The Subscription Manager agent connects the system to the Red Hat Customer Portal or to an on-premise subscription management service set-up by the customer using Subscription Asset Manager. Once connected, the customer can realize centralized control of their subscription assets and manage inventory, status, and gain enhanced reporting for the support of multiple systems. Scalability ----------- Scalability at the administrative level is increased by changes made to the Red Hat Enterprise Linux 6.5 beta kernel. Kernel dump files on large systems can now scale to multiple terabytes of data, and a new compression algorithm (LZO) speeds the creation of dump files, leading to reduced down time during crash dump generation and faster troubleshooting. An enhancement to the perf tool?s tracing and testing commands also provides additional infrastructure event monitoring capabilities. Networking ---------- Networking enhancements delivered in the beta of Red Hat Enterprise Linux 6.5 range from improved performance to expanded compliance with industry-specific requirements. These new capabilities include: * Better analysis of multicast traffic by inspecting Internet Group Management Protocol (IGMP) data to list router ports and groups with active subscribers * Stream Control Transmission Protocol (SCTP) association statistics in support of SS7 M3UA (Message Transfer Part Level 3 User Adaptation Layer) implementations provide additional monitoring capabilities for customers within the telecommunications sector * The introduction of IEEE 1588 Precision Time Protocol Version 2 (PTPv2), used for precisely synchronizing clocks in an Ethernet network. PTPv2 is capable of achieving clock accuracy in the sub-microsecond range and supports network driver time stamping for many of the widely adopted network drivers. Storage ------- The storage additions to the beta version of Red Hat Enterprise Linux 6.5 deliver improved scalability, performance, and ease of use. Enterprise storage customers benefit from improved control and recovery in iSCSI and Fibre Channel Storage Area Network (SAN) environments. The performance and high-availability features of multipath IO are available to a broader set of devices and multipath device automatic naming enhancements provide shorter, more convenient device names. The beta version of Red Hat Enterprise Linux 6.5 also supports Intel's NVM Express driver, the industry standard specification for accessing PCI Express bus-based SSDs. Also, FUSE-based file systems, like GlusterFS, can now use asynchronous IO for improved performance. Virtualization -------------- The beta version of Red Hat Enterprise Linux 6.5 continues to deliver virtualization to provide for a robust guest environment. The Red Hat guest can serve as a client to the highly scalable Gluster filesystem and Red Hat Storage. The path to Red Hat Enterprise Virtualization is streamlined by utilities that convert VMware OVF and Citrix Xen guests to Red Hat guests, helping customers to simplify their complex and diverse virtual infrastructure onto Kernel-based Virtual Machine (KVM) technology. The maximum memory for Red Hat guests has increased to four terabytes, allowing guests to run large-scale workloads, and dynamic hot-add functionality for virtual CPUs enables customers to add compute resources to installed guests on-the-fly, reducing downtime. Security -------- Security enhancements keep the beta version of Red Hat Enterprise Linux 6.5 up-to-date with essential government and industry regulations as well as emerging best practices. NSS, GnuTLS, OpenSSL and Java can now share a single, system-wide static data store, which can be used by crypto-toolkits as input for trusted certificate decisions, something that is required by many corporate deployments. Additional security capabilities include: * Support for smartcard authentication by providing single-sign-on capabilities across more application subsystems * The latest version of OpenSCAP, an authenticated scanner that meets the National Institute of Standards and Technology's (NIST) SCAP 1.2 certification requirements * Support for the latest Transport Layer Security (TLS 1.2) standard, allowing secure network communication. The beta of Red Hat Enterprise Linux 6.5 also features several desktop and graphical enhancements. Users can now remotely operate Windows 7 and Windows 8 desktops and Windows Server 2012 consoles using the RDP protocol, and customers running the Evolution mail client will now experience better integration with Microsoft Exchange, especially from a calendaring perspective. Additionally, the base LibreOffice software is updated to 4.0, which provides a host of office workflow improvements. Red Hat Enterprise Linux 6.5 beta is available now on Red Hat Network to all customers with an active Red Hat Enterprise Linux subscription. For access and downloads, please visit: https://rhn.redhat.com/rhn/software/channels/Beta.do We greatly appreciate the dedication and collaboration within the company and with our community to develop and deliver the highest quality open source enterprise platform available today. Sincerely, The Red Hat Enterprise Linux Team ADDITIONAL RESOURCES Please visit the resources below for more information about Red Hat Enterprise Linux 6.5. Access documentation for Red Hat Enterprise Linux 6.5 (requires login): * Red Hat Enterprise Linux 6.5 beta release notes: https://access.redhat.com/site/documentation/en-US/Red_Hat_Enterprise_Linux/6-Beta/html/6.5_Release_Notes/index.html * Red Hat Enterprise Linux 6.5 beta documentation: https://access.redhat.com/knowledge/docs/Red_Hat_Enterprise_Linux/ From bryan.hepworth at newcastle.ac.uk Thu Oct 10 21:17:35 2013 From: bryan.hepworth at newcastle.ac.uk (Bryan Hepworth) Date: Thu, 10 Oct 2013 21:17:35 +0000 Subject: [rhelv6-list] md and mdadm issue Message-ID: Hi all I recently had an issue with my md1200 with a Linux software raid. Originally I split it into 2 10TB partitions with Disk Manager. They received device numbers /dev/md126 and /dev/md127 - recently there was a bit of a power issue, and a DOA replacement drive, which has now been replaced. Unfortunately there now appears to be a couple of different device numbers, /dev/md0 and /dev/md1 and the new drive /dev/sdp as below: - unused devices: [root at inscribe2 mail]# lsscsi -g [0:0:32:0] enclosu DP BACKPLANE 1.07 - /dev/sg0 [0:2:0:0] disk DELL PERC H700 2.10 /dev/sda /dev/sg1 [0:2:1:0] disk DELL PERC H700 2.10 /dev/sdb /dev/sg2 [1:0:0:0] cd/dvd TEAC DVD-ROM DV-28SW R.2A /dev/sr0 /dev/sg3 [5:0:0:0] tape IBM ULT3580-HH4 A233 /dev/st0 /dev/sg4 [5:0:0:1] mediumx IBM 3573-TL 9.50 /dev/sch0 /dev/sg5 [5:0:1:0] disk WD WD2000FYYG D1B3 /dev/sdc /dev/sg6 [5:0:2:0] disk WD WD2000FYYG D1BC /dev/sdd /dev/sg7 [5:0:3:0] disk WD WD2000FYYG D1BC /dev/sde /dev/sg8 [5:0:4:0] disk WD WD2000FYYG D1BC /dev/sdf /dev/sg9 [5:0:5:0] disk WD WD2000FYYG D1B3 /dev/sdg /dev/sg10 [5:0:6:0] disk WD WD2000FYYG D1B3 /dev/sdh /dev/sg11 [5:0:7:0] disk WD WD2000FYYG D1B3 /dev/sdi /dev/sg12 [5:0:8:0] disk WD WD2000FYYG D1B3 /dev/sdj /dev/sg13 [5:0:9:0] disk WD WD2000FYYG D1B3 /dev/sdk /dev/sg14 [5:0:10:0] disk WD WD2000FYYG D1BB /dev/sdl /dev/sg15 [5:0:11:0] disk WD WD2000FYYG D1BB /dev/sdm /dev/sg16 [5:0:12:0] enclosu DELL MD1200 1.01 - /dev/sg17 [5:0:13:0] disk WD WD2001FYYG-01SL3 VR07 /dev/sdp /dev/sg22 [6:0:0:0] tape IBM ULT3580-HH4 A233 /dev/st1 /dev/sg18 [6:0:1:0] disk DELL MD32xx 0770 /dev/sdn /dev/sg19 [6:0:1:1] disk DELL MD32xx 0770 /dev/sdo /dev/sg20 [6:0:1:31] disk DELL Universal Xport 0770 - /dev/sg21 [root at inscribe2 mail]# cat /proc/mdstat Personalities : [raid6] [raid5] [raid4] md1 : inactive sdg1[0] 1953510977 blocks super 1.2 md0 : inactive sdf1[1] sdc1[6] sdd1[4] sde1[2] 7814043908 blocks super 1.2 md127 : active raid5 sdi1[4] sdh1[6] sdk1[2] sdj1[3] sdl1[1] sdm1[0] 9767552000 blocks super 1.2 level 5, 512k chunk, algorithm 2 [6/6] [UUUUUU] bitmap: 1/15 pages [4KB], 65536KB chunk unused devices: [root at inscribe2 mail]# How do I go about moving the drives back into /dev/md126 ? would I be right in thinking I issue mdadm --assemble --verbose /dev/md126 /dev/sdc /dev/sdd /dev/sde /dev/sdf /dev/sdg /dev/sdp Any pointers gratefully received on the process. Thanks Bryan Bryan Hepworth Computing Officer Institute of Genetic Medicine Newcastle University International Centre for Life Newcastle NE1 3BZ From gianluca.cecchi at gmail.com Tue Oct 15 08:26:46 2013 From: gianluca.cecchi at gmail.com (Gianluca Cecchi) Date: Tue, 15 Oct 2013 10:26:46 +0200 Subject: [rhelv6-list] Info on network install with 10Gbit/s Message-ID: Hello, I have a server through which I install using pxe and kickstart downloaded via network. The kickstart fie is downloaded via http. I have no problem installing both physical and virtual servers with 1Gb/s adapters. Now I'm trying on several types of servers with 10Gbit/s adapters and they successfully download via pxe the kernel and initrd, installation continues but seems to remain stuck in the step transferring kickstart file (it is a text install; I see the download feedback red line completed and this lets me think the file has been downloaded, I also see it in httpd access.log) Alt+F3 and Alt+F4 don't show anything interesting. No shell yet at Alt+F2 screen. I tried both with biosdevname=0 and without it but nothing changes. The OS tried is RH EL 6.4. Just to ask if there is any particular caveat for this kind of combination. I'm going to retrieve the hw model of the network adapters. At the moment I only know that they are using ixgbe kernel module. I dind't find anything special in 6.4 release notes. Only this information about version: " The ixgbe driver for Intel 10 Gigabit PCI Express network devices has been updated to version 3.9.15-k to include support for SR-IOV with Data Center Bridging (DCB) or Receive-Side Scaling (RSS), PT P support as a T echnology Preview, latest hardware support, enhancements, and bug fixes. " Thanks in advance, Gianluca From solarflow99 at gmail.com Tue Oct 15 13:37:03 2013 From: solarflow99 at gmail.com (solarflow99) Date: Tue, 15 Oct 2013 09:37:03 -0400 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: References: Message-ID: Is this with a virtual machine of a physical server? It seems like the initrd doesn't have the driver or the virtual nix driver isn't recognized. Depends what kind of hypervisor you are using. On Tue, Oct 15, 2013 at 4:26 AM, Gianluca Cecchi wrote: > Hello, > I have a server through which I install using pxe and kickstart > downloaded via network. > The kickstart fie is downloaded via http. > I have no problem installing both physical and virtual servers with > 1Gb/s adapters. > Now I'm trying on several types of servers with 10Gbit/s adapters and > they successfully download via pxe the kernel and initrd, installation > continues but seems to remain stuck in the step > > transferring kickstart file > > (it is a text install; I see the download feedback red line completed > and this lets me think the file has been downloaded, I also see it in > httpd access.log) > > Alt+F3 and Alt+F4 don't show anything interesting. > No shell yet at Alt+F2 screen. > I tried both with biosdevname=0 and without it but nothing changes. > The OS tried is RH EL 6.4. > > Just to ask if there is any particular caveat for this kind of combination. > I'm going to retrieve the hw model of the network adapters. At the > moment I only know that they are using ixgbe kernel module. > I dind't find anything special in 6.4 release notes. Only this > information about version: > " > The ixgbe driver for Intel 10 Gigabit PCI Express network devices has > been updated to version > 3.9.15-k to include support for SR-IOV with Data Center Bridging (DCB) > or Receive-Side Scaling > (RSS), PT P support as a T echnology Preview, latest hardware support, > enhancements, and bug > fixes. > " > Thanks in advance, > Gianluca > > _______________________________________________ > rhelv6-list mailing list > rhelv6-list at redhat.com > https://www.redhat.com/mailman/listinfo/rhelv6-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From gianluca.cecchi at gmail.com Tue Oct 15 13:43:08 2013 From: gianluca.cecchi at gmail.com (Gianluca Cecchi) Date: Tue, 15 Oct 2013 15:43:08 +0200 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: References: Message-ID: On Tue, Oct 15, 2013 at 3:37 PM, solarflow99 wrote: > Is this with a virtual machine of a physical server? It seems like the > initrd doesn't have the driver or the virtual nix driver isn't recognized. > Depends what kind of hypervisor you are using. This is on a physical sevrer that only has 10Gbit/s network adapters. Gianluca From Joerg.Mertin at ca.com Tue Oct 15 13:51:58 2013 From: Joerg.Mertin at ca.com (Mertin, Joerg) Date: Tue, 15 Oct 2013 13:51:58 +0000 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: References: Message-ID: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> Get a Driver Disk from the manufacturer of the network card, and install the RedHat system using dd (Tells it to use a driver disk). -- Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com -----Original Message----- From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Gianluca Cecchi Sent: Tuesday, October 15, 2013 3:43 PM To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s On Tue, Oct 15, 2013 at 3:37 PM, solarflow99 wrote: > Is this with a virtual machine of a physical server? It seems like > the initrd doesn't have the driver or the virtual nix driver isn't recognized. > Depends what kind of hypervisor you are using. This is on a physical sevrer that only has 10Gbit/s network adapters. Gianluca _______________________________________________ rhelv6-list mailing list rhelv6-list at redhat.com https://www.redhat.com/mailman/listinfo/rhelv6-list From gianluca.cecchi at gmail.com Tue Oct 15 14:02:24 2013 From: gianluca.cecchi at gmail.com (Gianluca Cecchi) Date: Tue, 15 Oct 2013 16:02:24 +0200 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> Message-ID: On Tue, Oct 15, 2013 at 3:51 PM, Mertin, Joerg wrote: > Get a Driver Disk from the manufacturer of the network card, and install the RedHat system using dd (Tells it to use a driver disk). Yes, I was thinking abut it too, as the Intel ixgbe driver seems far ahead, at least in version.... Never tried to embed use of dd in automated install via kickstart, so I don't know if it is possible and how to do it... Any tip welcome Gianluca From F.REYNIER at cr-poitou-charentes.fr Tue Oct 15 14:03:23 2013 From: F.REYNIER at cr-poitou-charentes.fr (F.REYNIER at cr-poitou-charentes.fr) Date: Tue, 15 Oct 2013 16:03:23 +0200 Subject: [rhelv6-list] FLORENT REYNIER est absent(e). Message-ID: Je serai absent(e) du Lun 14/10/2013 au Ven 18/10/2013. Je r?pondrai ? votre message d?s mon retour. Pour connaitre les actions de la R?gion Poitou-Charentes et r?aliser vos projets, rendez-vous sur le site www.poitou-charentes.fr From stan.hearn at nscorp.com Tue Oct 15 14:05:45 2013 From: stan.hearn at nscorp.com (Hearn, Stan J.) Date: Tue, 15 Oct 2013 10:05:45 -0400 Subject: [rhelv6-list] Entitlements as it relates to sockets Message-ID: What are you being told about sockets and entitlements. If a system has four sockets with only two populated CPUs, you must purchase the "Up to 4" socket entitlements? Is the answer different between RHEL 5 and RHEL 6? Is the answer different when purchasing entitlements from HP or Red Hat? Thanks, Stan From Joerg.Mertin at ca.com Tue Oct 15 14:07:04 2013 From: Joerg.Mertin at ca.com (Mertin, Joerg) Date: Tue, 15 Oct 2013 14:07:04 +0000 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> Message-ID: <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> It is possible - I do it on occasions (I created some 20 Kickstart images myself in the last 2 years). The dd command is only "added" to the current kickstart action and asks then for a driver disk before continuing the installation. The only part where it fails is when the kernel gets updated and does not provide the driver :} There, you will have to reboot into the old kernel, get the new driver and install it manually for the new kernel release. -- Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com -----Original Message----- From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Gianluca Cecchi Sent: Tuesday, October 15, 2013 4:02 PM To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s On Tue, Oct 15, 2013 at 3:51 PM, Mertin, Joerg wrote: > Get a Driver Disk from the manufacturer of the network card, and install the RedHat system using dd (Tells it to use a driver disk). Yes, I was thinking abut it too, as the Intel ixgbe driver seems far ahead, at least in version.... Never tried to embed use of dd in automated install via kickstart, so I don't know if it is possible and how to do it... Any tip welcome Gianluca _______________________________________________ rhelv6-list mailing list rhelv6-list at redhat.com https://www.redhat.com/mailman/listinfo/rhelv6-list From jean-yves at lenhof.eu.org Tue Oct 15 14:20:17 2013 From: jean-yves at lenhof.eu.org (jean-yves at lenhof.eu.org) Date: Tue, 15 Oct 2013 16:20:17 +0200 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> Message-ID: <24d07d521902f202a7dfc196daefd35a@lenhof.eu.org> Le 2013-10-15 16:02, Gianluca Cecchi a ?crit?: > On Tue, Oct 15, 2013 at 3:51 PM, Mertin, Joerg wrote: >> Get a Driver Disk from the manufacturer of the network card, and >> install the RedHat system using dd (Tells it to use a driver disk). > > Yes, I was thinking abut it too, as the Intel ixgbe driver seems far > ahead, at least in version.... > Never tried to embed use of dd in automated install via kickstart, so > I don't know if it is possible and how to do it... > Any tip welcome Hi, driverdisk seems to be the keyword you're looking for : https://access.redhat.com/site/documentation/fr-FR/Red_Hat_Enterprise_Linux/6/html/Installation_Guide/s1-kickstart2-options.html You'll probably need to copy the driver on a local disk (unless the network can be started even if not perfect to access via htttp the driverdisk) Regards, JYL From wolfy at nobugconsulting.ro Tue Oct 15 14:28:52 2013 From: wolfy at nobugconsulting.ro (Manuel Wolfshant) Date: Tue, 15 Oct 2013 17:28:52 +0300 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> Message-ID: <525D5124.4030500@nobugconsulting.ro> On 10/15/2013 05:07 PM, Mertin, Joerg wrote: > It is possible - I do it on occasions (I created some 20 Kickstart images myself in the last 2 years). > The dd command is only "added" to the current kickstart action and asks then for a driver disk before continuing the installation. > The only part where it fails is when the kernel gets updated and does not provide the driver :} There, you will have to reboot into the old kernel, get the new driver and install it manually for the new kernel release. http://elrepo.org/tiki/kmod-ixgbe > > -- > Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com > > > -----Original Message----- > From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Gianluca Cecchi > Sent: Tuesday, October 15, 2013 4:02 PM > To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list > Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s > > On Tue, Oct 15, 2013 at 3:51 PM, Mertin, Joerg wrote: >> Get a Driver Disk from the manufacturer of the network card, and install the RedHat system using dd (Tells it to use a driver disk). > Yes, I was thinking abut it too, as the Intel ixgbe driver seems far ahead, at least in version.... > Never tried to embed use of dd in automated install via kickstart, so I don't know if it is possible and how to do it... > Any tip welcome > > Gianluca From Joerg.Mertin at ca.com Tue Oct 15 14:59:09 2013 From: Joerg.Mertin at ca.com (Mertin, Joerg) Date: Tue, 15 Oct 2013 14:59:09 +0000 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: <525D5124.4030500@nobugconsulting.ro> References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> <525D5124.4030500@nobugconsulting.ro> Message-ID: <34F5E86C3ADB304DA6D20B688F0E9BB17F045206@usilms111b.ca.com> Yup - I know that. Unfortunately, from experience, this does not always work - especially when having security fixes which change the ABI. Well - on my workstation using Ubuntu that happens quite often. Using RedHat only on the servers... -- Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com -----Original Message----- From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Manuel Wolfshant Sent: Tuesday, October 15, 2013 4:29 PM To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s On 10/15/2013 05:07 PM, Mertin, Joerg wrote: > It is possible - I do it on occasions (I created some 20 Kickstart images myself in the last 2 years). > The dd command is only "added" to the current kickstart action and asks then for a driver disk before continuing the installation. > The only part where it fails is when the kernel gets updated and does not provide the driver :} There, you will have to reboot into the old kernel, get the new driver and install it manually for the new kernel release. http://elrepo.org/tiki/kmod-ixgbe > > -- > Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com > > > -----Original Message----- > From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Gianluca Cecchi > Sent: Tuesday, October 15, 2013 4:02 PM > To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list > Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s > > On Tue, Oct 15, 2013 at 3:51 PM, Mertin, Joerg wrote: >> Get a Driver Disk from the manufacturer of the network card, and install the RedHat system using dd (Tells it to use a driver disk). > Yes, I was thinking abut it too, as the Intel ixgbe driver seems far ahead, at least in version.... > Never tried to embed use of dd in automated install via kickstart, so I don't know if it is possible and how to do it... > Any tip welcome > > Gianluca _______________________________________________ rhelv6-list mailing list rhelv6-list at redhat.com https://www.redhat.com/mailman/listinfo/rhelv6-list From wolfy at nobugconsulting.ro Tue Oct 15 15:15:34 2013 From: wolfy at nobugconsulting.ro (Manuel Wolfshant) Date: Tue, 15 Oct 2013 18:15:34 +0300 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: <34F5E86C3ADB304DA6D20B688F0E9BB17F045206@usilms111b.ca.com> References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> <525D5124.4030500@nobugconsulting.ro> <34F5E86C3ADB304DA6D20B688F0E9BB17F045206@usilms111b.ca.com> Message-ID: <525D5C16.3050400@nobugconsulting.ro> On 10/15/2013 05:59 PM, Mertin, Joerg wrote: > Yup - I know that. > Unfortunately, from experience, this does not always work - especially when having security fixes which change the ABI. Well - on my workstation using Ubuntu that happens quite often. > Using RedHat only on the servers... ABI changes extremely rarely in RH. And elrepo's kmods are created so as to take advantage of the weak-modules mechanism so they can be reused without issues after kernel upgrades. Which is the core point of the whole story And kmod packages can be integrated in driver disks with a tiny bit of work. > > Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com > > > -----Original Message----- > From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Manuel Wolfshant > Sent: Tuesday, October 15, 2013 4:29 PM > To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list > Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s > > On 10/15/2013 05:07 PM, Mertin, Joerg wrote: >> It is possible - I do it on occasions (I created some 20 Kickstart images myself in the last 2 years). >> The dd command is only "added" to the current kickstart action and asks then for a driver disk before continuing the installation. >> The only part where it fails is when the kernel gets updated and does not provide the driver :} There, you will have to reboot into the old kernel, get the new driver and install it manually for the new kernel release. > > http://elrepo.org/tiki/kmod-ixgbe > > > >> -- >> Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com >> >> >> -----Original Message----- >> From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Gianluca Cecchi >> Sent: Tuesday, October 15, 2013 4:02 PM >> To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list >> Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s >> >> On Tue, Oct 15, 2013 at 3:51 PM, Mertin, Joerg wrote: >>> Get a Driver Disk from the manufacturer of the network card, and install the RedHat system using dd (Tells it to use a driver disk). >> Yes, I was thinking abut it too, as the Intel ixgbe driver seems far ahead, at least in version.... >> Never tried to embed use of dd in automated install via kickstart, so I don't know if it is possible and how to do it... >> Any tip welcome >> >> Gianluca From Joerg.Mertin at ca.com Tue Oct 15 15:22:13 2013 From: Joerg.Mertin at ca.com (Mertin, Joerg) Date: Tue, 15 Oct 2013 15:22:13 +0000 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: <525D5C16.3050400@nobugconsulting.ro> References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> <525D5124.4030500@nobugconsulting.ro> <34F5E86C3ADB304DA6D20B688F0E9BB17F045206@usilms111b.ca.com> <525D5C16.3050400@nobugconsulting.ro> Message-ID: <34F5E86C3ADB304DA6D20B688F0E9BB17F04522C@usilms111b.ca.com> I fully agree on that :) and like it, however I had to deal with some ABI changes while 15K Km away, and that is not always easy when the remote party does not know what a CLI is ... -- Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com -----Original Message----- From: rhelv6-list-bounces at redhat.com [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Manuel Wolfshant Sent: Tuesday, October 15, 2013 5:16 PM To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s On 10/15/2013 05:59 PM, Mertin, Joerg wrote: > Yup - I know that. > Unfortunately, from experience, this does not always work - especially when having security fixes which change the ABI. Well - on my workstation using Ubuntu that happens quite often. > Using RedHat only on the servers... ABI changes extremely rarely in RH. And elrepo's kmods are created so as to take advantage of the weak-modules mechanism so they can be reused without issues after kernel upgrades. Which is the core point of the whole story And kmod packages can be integrated in driver disks with a tiny bit of work. > > Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | > APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | joerg.mertin at ca.com > > > -----Original Message----- > From: rhelv6-list-bounces at redhat.com > [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Manuel Wolfshant > Sent: Tuesday, October 15, 2013 4:29 PM > To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list > Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s > > On 10/15/2013 05:07 PM, Mertin, Joerg wrote: >> It is possible - I do it on occasions (I created some 20 Kickstart images myself in the last 2 years). >> The dd command is only "added" to the current kickstart action and asks then for a driver disk before continuing the installation. >> The only part where it fails is when the kernel gets updated and does not provide the driver :} There, you will have to reboot into the old kernel, get the new driver and install it manually for the new kernel release. > > http://elrepo.org/tiki/kmod-ixgbe > > > >> -- >> Joerg Mertin | Sr. Engineering Services Architect | CA Technologies | >> APM Global SWAT Team | Mobile: +33 6 19 51 90 24 | >> joerg.mertin at ca.com >> >> >> -----Original Message----- >> From: rhelv6-list-bounces at redhat.com >> [mailto:rhelv6-list-bounces at redhat.com] On Behalf Of Gianluca Cecchi >> Sent: Tuesday, October 15, 2013 4:02 PM >> To: Red Hat Enterprise Linux 6 (Santiago) discussion mailing-list >> Subject: Re: [rhelv6-list] Info on network install with 10Gbit/s >> >> On Tue, Oct 15, 2013 at 3:51 PM, Mertin, Joerg wrote: >>> Get a Driver Disk from the manufacturer of the network card, and install the RedHat system using dd (Tells it to use a driver disk). >> Yes, I was thinking abut it too, as the Intel ixgbe driver seems far ahead, at least in version.... >> Never tried to embed use of dd in automated install via kickstart, so I don't know if it is possible and how to do it... >> Any tip welcome >> >> Gianluca _______________________________________________ rhelv6-list mailing list rhelv6-list at redhat.com https://www.redhat.com/mailman/listinfo/rhelv6-list From solarflow99 at gmail.com Tue Oct 15 15:23:56 2013 From: solarflow99 at gmail.com (solarflow99) Date: Tue, 15 Oct 2013 11:23:56 -0400 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: <525D5C16.3050400@nobugconsulting.ro> References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> <525D5124.4030500@nobugconsulting.ro> <34F5E86C3ADB304DA6D20B688F0E9BB17F045206@usilms111b.ca.com> <525D5C16.3050400@nobugconsulting.ro> Message-ID: Hasn't this changed a while back? I thought there is a setting that has to be disabled to allow a module to be introduced with an updated kernel with a newer revision number. mechanism so they can be reused without issues after kernel upgrades. Which is the core point of the whole storyAnd kmod packages can be integrated in driver disks with a tiny bit of work. -------------- next part -------------- An HTML attachment was scrubbed... URL: From amyagi at gmail.com Tue Oct 15 21:50:24 2013 From: amyagi at gmail.com (Akemi Yagi) Date: Tue, 15 Oct 2013 14:50:24 -0700 Subject: [rhelv6-list] Info on network install with 10Gbit/s In-Reply-To: <34F5E86C3ADB304DA6D20B688F0E9BB17F04522C@usilms111b.ca.com> References: <34F5E86C3ADB304DA6D20B688F0E9BB17F045155@usilms111b.ca.com> <34F5E86C3ADB304DA6D20B688F0E9BB17F0451BD@usilms111b.ca.com> <525D5124.4030500@nobugconsulting.ro> <34F5E86C3ADB304DA6D20B688F0E9BB17F045206@usilms111b.ca.com> <525D5C16.3050400@nobugconsulting.ro> <34F5E86C3ADB304DA6D20B688F0E9BB17F04522C@usilms111b.ca.com> Message-ID: On Tue, Oct 15, 2013 at 8:22 AM, Mertin, Joerg wrote: > I fully agree on that :) and like it, however I had to deal with some ABI changes while 15K Km away, and that is not always easy when the remote party does not know what a CLI is ... Yes, kABI changes do happen, but very rarely as Manuel said. ELRepo addresses this by rebuilding the packages that are affected. Please try filing a request if you find any kmod that requires such updates ( http://elrepo.org/bugs ). Akemi From linux at cmadams.net Tue Oct 22 17:34:27 2013 From: linux at cmadams.net (Chris Adams) Date: Tue, 22 Oct 2013 12:34:27 -0500 Subject: [rhelv6-list] NFS ID map? Message-ID: <20131022173427.GB7552@cmadams.net> I've got some systems that I'm replacing, and while migrating data from the old NFS servers to the new NFS servers, I ran into a dumb mistake: the user/group IDs don't match between the old and new (I thought I was going to be able to shut down the old services, sync the data, and bring up the new, so it didn't matter then, but that isn't going to work). I know NFS has some type of ID mapping, but I've never messed with it. Is this something that it can help with? I have data owned by 500/500, accessed by a service running as 500/500, but the new NFS server ended up with the user as 1000/1000 (and I need to mount the new storage on the old servers for migration). Any suggestions, or am I stuck with reconfiguring all the new stuff to use the old user/group IDs? -- Chris Adams From b.j.smith at ieee.org Tue Oct 22 18:08:13 2013 From: b.j.smith at ieee.org (Bryan J Smith) Date: Tue, 22 Oct 2013 14:08:13 -0400 Subject: [rhelv6-list] NFS ID map? In-Reply-To: <20131022173427.GB7552@cmadams.net> References: <20131022173427.GB7552@cmadams.net> Message-ID: NFS4 uses rpc.idmapd, not longer using UID/GID. So as long as the user and group names match, you don't have to worry about UID/GID at all. One thing to keep in-mind is that it's actually more than just the names. It's actually a principal -- i.e., user at REALM. So either you have to use Kerberos, or set the appropriate Domain (defaults to domain name of system) and/or Local-Realm (equivalent aliases) in the file idmapd.conf itself. On Tue, Oct 22, 2013 at 1:34 PM, Chris Adams wrote: > I've got some systems that I'm replacing, and while migrating data from > the old NFS servers to the new NFS servers, I ran into a dumb mistake: > the user/group IDs don't match between the old and new (I thought I was > going to be able to shut down the old services, sync the data, and bring > up the new, so it didn't matter then, but that isn't going to work). > > I know NFS has some type of ID mapping, but I've never messed with it. > Is this something that it can help with? > > I have data owned by 500/500, accessed by a service running as 500/500, > but the new NFS server ended up with the user as 1000/1000 (and I need > to mount the new storage on the old servers for migration). > > Any suggestions, or am I stuck with reconfiguring all the new stuff to > use the old user/group IDs? > -- > Chris Adams > > _______________________________________________ > rhelv6-list mailing list > rhelv6-list at redhat.com > https://www.redhat.com/mailman/listinfo/rhelv6-list -- -- Bryan J Smith - Professional, Technical Annoyance b.j.smith at ieee.org - http://www.linkedin.com/in/bjsmith From linux at cmadams.net Tue Oct 22 18:39:14 2013 From: linux at cmadams.net (Chris Adams) Date: Tue, 22 Oct 2013 13:39:14 -0500 Subject: [rhelv6-list] NFS ID map? In-Reply-To: References: <20131022173427.GB7552@cmadams.net> Message-ID: <20131022183914.GC7552@cmadams.net> Once upon a time, Bryan J Smith said: > NFS4 uses rpc.idmapd, not longer using UID/GID. So as long as the > user and group names match, you don't have to worry about UID/GID at > all. Okay, that sounds good. > One thing to keep in-mind is that it's actually more than just the > names. It's actually a principal -- i.e., user at REALM. So either you > have to use Kerberos, or set the appropriate Domain (defaults to > domain name of system) and/or Local-Realm (equivalent aliases) in the > file idmapd.conf itself. Okay, thanks. I've never messed with NFS ID mapping before (always just made sure everything matched). However, a new problem: the existing mounts are NFSv3. I can get one of the new (RHEL 6) servers to access the share via NFSv4, but I can't get an NFSv4 mount to work on the old server (running RHEL 5). I added the export with a call to exportfs like: # exportfs -i -o sync,rw,no_root_squash x.x.x.x/x:/srv/foo The RHEL 6 clients can mount -t nfs4 x.x.x.x:/foo, but the RHEL 5 clients get a "No such file or directory" for a "-t nfs4" mount (I tried both /foo and /srv/foo). NFSv3 works just fine. From what I understand, NFSv4 mounts are handled in the kernel, rather than with rpc.mountd, so maybe I'm not getting something updated right? -- Chris Adams From b.j.smith at ieee.org Tue Oct 22 18:51:51 2013 From: b.j.smith at ieee.org (Bryan J Smith) Date: Tue, 22 Oct 2013 14:51:51 -0400 Subject: [rhelv6-list] NFS ID map? In-Reply-To: <20131022183914.GC7552@cmadams.net> References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> Message-ID: Both RHEL5 and RHEL6 work fine with NFS4. Although RHEL4 is capable of supporting NFS4, I don't recommend it. So ... are you sure you're actually getting NFS4 on RHEL6? Check with /proc/mounts to be sure. ;) Long story short ... file system type "nfs" during mount differs on RHEL6 from RHEL5. You use "nfs" with option "vers=4" and not just type "nfs4" on RHEL6. In fact, although don't quote me, but I think nfs4 = nfs on RHEL6, with the "vers=" being everything. Another reason I say this ... NFS4 exports work completely _different_ than NFSv3. You _must_ use fsid=0 and mount on the server under that tree. That's completely different than NFSv3. One can use bind mounts to solve such, but it also offers the nice feature of no longer having to export everything. Here's an example /etc/export from one of my servers. Note the differences in not just fsid, but "subtree_check,crossmnt" (NFS4) and "no_subtreecheck" (NFSv3). # # /etc/exports # # NFS4 fsid=0 # - First line is read-write w/full privacy (auth/auth/crypt/sign), # but requires gss/krb5 (Kerberos) # - Second line trusts a subnet (a.b.c.d/cidr) for read-only, # but the idmapd domains/realm aliases must still match /exports gss/krb5p(rw,insecure,no_root_squash,subtree_check,crossmnt,fsid=0) \ 172.16.x.x/20(ro,insecure,root_squash,subtree_check,crossmnt,fsid=0) # # NFSv3 fsid=3xxx # - Legacy exports for NFSv3, one per share, read-only /exports/share1 172.16.x.x/20(ro,insecure,no_subtree_check,no_root_squash,nohide,fsid=3010) /exports/share2 172.16.x.x/20(ro,insecure,no_subtree_check,no_root_squash,nohide,fsid=3020) /exports/share3 172.16.x.x/20(ro,insecure,no_subtree_check,no_root_squash,nohide,fsid=3030) -- Bryan J Smith - Professional, Technical Annoyance b.j.smith at ieee.org - http://www.linkedin.com/in/bjsmith From linux at cmadams.net Tue Oct 22 19:02:58 2013 From: linux at cmadams.net (Chris Adams) Date: Tue, 22 Oct 2013 14:02:58 -0500 Subject: [rhelv6-list] NFS ID map? In-Reply-To: References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> Message-ID: <20131022190258.GD7552@cmadams.net> Once upon a time, Bryan J Smith said: > Both RHEL5 and RHEL6 work fine with NFS4. Although RHEL4 is capable > of supporting NFS4, I don't recommend it. > > So ... are you sure you're actually getting NFS4 on RHEL6? Check with > /proc/mounts to be sure. ;) > > Long story short ... file system type "nfs" during mount differs on > RHEL6 from RHEL5. You use "nfs" with option "vers=4" and not just > type "nfs4" on RHEL6. In fact, although don't quote me, but I think > nfs4 = nfs on RHEL6, with the "vers=" being everything. Yep, I did check /proc/mounts to make sure. I can mount /srv/foo with nfsvers=3 and /foo with nfsvers=4 on RHEL 6 (-t nfs4 also does still get nfsvers=4 on RHEL 6; I think that's gone away with newer Fedoras, so probably will in RHEL 7). > Another reason I say this ... NFS4 exports work completely _different_ > than NFSv3. You _must_ use fsid=0 and mount on the server under that > tree. That's completely different than NFSv3. One can use bind > mounts to solve such, but it also offers the nice feature of no longer > having to export everything. Yes, I have /srv with fsid=0. The RHEL 6 clients can definately mount with both NFSv3 and NFSv4 (checking /proc/mounts after mounting), but the RHEL 5 clients are only working with NFSv3 (when I need NFSv4 for the ID map). Oh, I figured out my problem - this temporary/migration export is on a different subnet, and my fsid=0 export was was only for the new subnet. D'oh! Fixed that, and NFSv4, and ID mapping (with the default config), are now working. Now, on with the (migration) show! Thanks! -- Chris Adams From linux at cmadams.net Tue Oct 22 19:19:49 2013 From: linux at cmadams.net (Chris Adams) Date: Tue, 22 Oct 2013 14:19:49 -0500 Subject: [rhelv6-list] NFS ID map? In-Reply-To: <20131022190258.GD7552@cmadams.net> References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> <20131022190258.GD7552@cmadams.net> Message-ID: <20131022191949.GE7552@cmadams.net> Once upon a time, Chris Adams said: > Oh, I figured out my problem - this temporary/migration export is on a > different subnet, and my fsid=0 export was was only for the new subnet. > D'oh! Fixed that, and NFSv4, and ID mapping (with the default config), > are now working. Now, on with the (migration) show! Well, I thought ID mapping was working. "ls" shows the right user, but I can't actually access directories as that user on the NFS client: $ id uid=500(vmail) gid=500(vmail) groups=500(vmail) $ ls -ld tmp drwx------ 2 vmail vmail 4096 Oct 22 14:09 tmp $ ls -l tmp ls: tmp: Permission denied The vmail user/group is 1000 on the NFS server. Why would the translation work for stat() calls from ls, but not when I try to actually access the files? -- Chris Adams From jlb17 at duke.edu Tue Oct 22 19:29:40 2013 From: jlb17 at duke.edu (Joshua Baker-LePain) Date: Tue, 22 Oct 2013 12:29:40 -0700 (PDT) Subject: [rhelv6-list] NFS ID map? In-Reply-To: <20131022191949.GE7552@cmadams.net> References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> <20131022190258.GD7552@cmadams.net> <20131022191949.GE7552@cmadams.net> Message-ID: On Tue, 22 Oct 2013 at 12:19pm, Chris Adams wrote > Once upon a time, Chris Adams said: >> Oh, I figured out my problem - this temporary/migration export is on a >> different subnet, and my fsid=0 export was was only for the new subnet. >> D'oh! Fixed that, and NFSv4, and ID mapping (with the default config), >> are now working. Now, on with the (migration) show! > > Well, I thought ID mapping was working. "ls" shows the right user, but > I can't actually access directories as that user on the NFS client: > > $ id > uid=500(vmail) gid=500(vmail) groups=500(vmail) > $ ls -ld tmp > drwx------ 2 vmail vmail 4096 Oct 22 14:09 tmp > $ ls -l tmp > ls: tmp: Permission denied > > The vmail user/group is 1000 on the NFS server. Why would the > translation work for stat() calls from ls, but not when I try to > actually access the files? Are you using Kerberos with NFSv4? If not, then the numeric UIDs and GIDs still must match. idmapd *will* map usernames for display purposes (i.e. what 'ls -l' sees), but the numeric IDs are still used for permissions. Clear as mud, ain't it? At least, that's my experience of it... -- Joshua Baker-LePain QB3 Shared Cluster Sysadmin UCSF From linux at cmadams.net Tue Oct 22 19:35:17 2013 From: linux at cmadams.net (Chris Adams) Date: Tue, 22 Oct 2013 14:35:17 -0500 Subject: [rhelv6-list] NFS ID map? In-Reply-To: References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> <20131022190258.GD7552@cmadams.net> <20131022191949.GE7552@cmadams.net> Message-ID: <20131022193517.GF7552@cmadams.net> Once upon a time, Joshua Baker-LePain said: > Are you using Kerberos with NFSv4? If not, then the numeric UIDs > and GIDs still must match. idmapd *will* map usernames for display > purposes (i.e. what 'ls -l' sees), but the numeric IDs are still > used for permissions. Clear as mud, ain't it? Oh, that may be why I've always been confused by the ID mapping (and avoided it). Now, I'm not using Kerberos, so it sounds like the ID mapping won't actually help me with my problem. What's the point of ID mapping if it only maps them in some places? That's terribly confusing. stat() sees the numeric ID mapped (and chown()), but I guess open() permission checks don't? -- Chris Adams From jlb17 at duke.edu Tue Oct 22 19:44:27 2013 From: jlb17 at duke.edu (Joshua Baker-LePain) Date: Tue, 22 Oct 2013 12:44:27 -0700 (PDT) Subject: [rhelv6-list] NFS ID map? In-Reply-To: <20131022193517.GF7552@cmadams.net> References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> <20131022190258.GD7552@cmadams.net> <20131022191949.GE7552@cmadams.net> <20131022193517.GF7552@cmadams.net> Message-ID: On Tue, 22 Oct 2013 at 12:35pm, Chris Adams wrote > Once upon a time, Joshua Baker-LePain said: >> Are you using Kerberos with NFSv4? If not, then the numeric UIDs >> and GIDs still must match. idmapd *will* map usernames for display >> purposes (i.e. what 'ls -l' sees), but the numeric IDs are still >> used for permissions. Clear as mud, ain't it? > > Oh, that may be why I've always been confused by the ID mapping (and > avoided it). Now, I'm not using Kerberos, so it sounds like the ID > mapping won't actually help me with my problem. > > What's the point of ID mapping if it only maps them in some places? > That's terribly confusing. Yes. Yes it is. > stat() sees the numeric ID mapped (and chown()), but I guess open() > permission checks don't? This (rather old) thread explains it better than I can: http://thread.gmane.org/gmane.linux.nfsv4/7103/focus=7105 -- Joshua Baker-LePain QB3 Shared Cluster Sysadmin UCSF From F.REYNIER at cr-poitou-charentes.fr Tue Oct 22 20:03:41 2013 From: F.REYNIER at cr-poitou-charentes.fr (F.REYNIER at cr-poitou-charentes.fr) Date: Tue, 22 Oct 2013 22:03:41 +0200 Subject: [rhelv6-list] FLORENT REYNIER est absent(e). Message-ID: Je serai absent(e) du Lun 14/10/2013 au Mer 23/10/2013. Je r?pondrai ? votre message d?s mon retour. Pour connaitre les actions de la R?gion Poitou-Charentes et r?aliser vos projets, rendez-vous sur le site www.poitou-charentes.fr From b.j.smith at ieee.org Tue Oct 22 20:13:22 2013 From: b.j.smith at ieee.org (Bryan J Smith) Date: Tue, 22 Oct 2013 16:13:22 -0400 Subject: [rhelv6-list] NFS ID map? In-Reply-To: References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> <20131022190258.GD7552@cmadams.net> <20131022191949.GE7552@cmadams.net> <20131022193517.GF7552@cmadams.net> Message-ID: Oh man, you're kidding me?! What is the purpose of Domain/Local-Realms then? Sigh ... On Oct 22, 2013 3:45 PM, "Joshua Baker-LePain" wrote: > On Tue, 22 Oct 2013 at 12:35pm, Chris Adams wrote > > Once upon a time, Joshua Baker-LePain said: >> >>> Are you using Kerberos with NFSv4? If not, then the numeric UIDs >>> and GIDs still must match. idmapd *will* map usernames for display >>> purposes (i.e. what 'ls -l' sees), but the numeric IDs are still >>> used for permissions. Clear as mud, ain't it? >>> >> >> Oh, that may be why I've always been confused by the ID mapping (and >> avoided it). Now, I'm not using Kerberos, so it sounds like the ID >> mapping won't actually help me with my problem. >> >> What's the point of ID mapping if it only maps them in some places? >> That's terribly confusing. >> > > Yes. Yes it is. > > stat() sees the numeric ID mapped (and chown()), but I guess open() >> permission checks don't? >> > > This (rather old) thread explains it better than I can: > > http://thread.gmane.org/gmane.**linux.nfsv4/7103/focus=7105 > > -- > Joshua Baker-LePain > QB3 Shared Cluster Sysadmin > UCSF > > ______________________________**_________________ > rhelv6-list mailing list > rhelv6-list at redhat.com > https://www.redhat.com/**mailman/listinfo/rhelv6-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From linux at cmadams.net Tue Oct 22 20:31:43 2013 From: linux at cmadams.net (Chris Adams) Date: Tue, 22 Oct 2013 15:31:43 -0500 Subject: [rhelv6-list] NFS ID map? In-Reply-To: References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> <20131022190258.GD7552@cmadams.net> <20131022191949.GE7552@cmadams.net> <20131022193517.GF7552@cmadams.net> Message-ID: <20131022203143.GG7552@cmadams.net> Once upon a time, Bryan J Smith said: > Oh man, you're kidding me?! > > What is the purpose of Domain/Local-Realms then? Sigh ... I guess it is there to confuse the NFS-unfamiliar like me. :( I don't understand why somebody would take the time to write code and documentation that can't really ever be useful (or at least, I don't see where having stat() and open() have different results would be useful, except maybe to test the new junior sysadmin, or see which programmer thinks the access() system call should ever be used). Thanks for everybody's help and pointers, I guess I'm back to chowning a terabyte of Maildir... -- Chris Adams From b.j.smith at ieee.org Tue Oct 22 21:06:47 2013 From: b.j.smith at ieee.org (Bryan J Smith) Date: Tue, 22 Oct 2013 17:06:47 -0400 Subject: [rhelv6-list] NFS ID map? In-Reply-To: <20131022203143.GG7552@cmadams.net> References: <20131022173427.GB7552@cmadams.net> <20131022183914.GC7552@cmadams.net> <20131022190258.GD7552@cmadams.net> <20131022191949.GE7552@cmadams.net> <20131022193517.GF7552@cmadams.net> <20131022203143.GG7552@cmadams.net> Message-ID: I've just gotten lucky I guess. I've either, always had Kerberos (e.g., IdM), or the UIDs must have matched. On Oct 22, 2013 4:33 PM, "Chris Adams" wrote: > Once upon a time, Bryan J Smith said: > > Oh man, you're kidding me?! > > > > What is the purpose of Domain/Local-Realms then? Sigh ... > > I guess it is there to confuse the NFS-unfamiliar like me. :( > > I don't understand why somebody would take the time to write code and > documentation that can't really ever be useful (or at least, I don't see > where having stat() and open() have different results would be useful, > except maybe to test the new junior sysadmin, or see which programmer > thinks the access() system call should ever be used). > > Thanks for everybody's help and pointers, I guess I'm back to chowning a > terabyte of Maildir... > > -- > Chris Adams > > _______________________________________________ > rhelv6-list mailing list > rhelv6-list at redhat.com > https://www.redhat.com/mailman/listinfo/rhelv6-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: