[Freeipa-users] ns-slapd hangs for 2-3 minutes, then resumes.

Mon Jun 13 22:30:59 UTC 2016

On 06/13/2016 01:13 PM, Guillermo Fuentes wrote:
> Hi Rich,
>
> After I started running the stack traces, the problem hasn't happen as
> frequently as it use to but today I was able to get the stack traces.
> As they aren't similar I'll send them over to you in a separate email.
>
> This is what I did to start the stack traces (CentOS 7):
> # yum install -y --enablerepo=base-debuginfo 389-ds-base-debuginfo
> ipa-debuginfo slapi-nis-debuginfo nspr-debuginfo
> # yum install -y gdb
> # systemctl stop ipa.service ; sleep 10; systemctl start ipa.service
> # mkdir -p /var/log/stacktraces
>
> Setup crontab to run the following every minute:
> gdb -ex 'set confirm off' -ex 'set pagination off' -ex 'thread apply
> all bt full' -ex 'quit' /usr/sbin/ns-slapd `pidof ns-slapd` >
> /var/log/stacktraces/stacktrace.`date +%s`.txt 2>&1

It looks similar to https://fedorahosted.org/389/ticket/48341 but you 
already have that fix.

One of the problems is that ids_sasl_check_bind acquires the connection 
lock and holds it for a very long time, which causes the main loop to 
block on that connection, which is similar to the above problem, and 
also similar to https://fedorahosted.org/389/ticket/48882.  Basically, 
anything which holds the connection c_mutex lock too long can hang the 
server.  In your case, this stack trace:

poll sss_cli_make_request_nochecks sss_cli_check_socket 
sss_pac_make_request sssdpac_verify krb5int_authdata_verify 
rd_req_decoded_opt krb5_rd_req_decoded kg_accept_krb5 
krb5_gss_accept_sec_context_ext krb5_gss_accept_sec_context 
gss_accept_sec_context gssapi_server_mech_step sasl_server_step 
sasl_server_start ids_sasl_check_bind do_bind 
connection_dispatch_operation _pt_root start_thread clone

I'm not sure if this particular situation is known/fixed.  Perhaps there 
is a way to make the poll() called by sss_cli_make_request_nochecks() 
have a smaller timeout?

Does this look familiar to any ipa/sssd developer?

>
> Thank you so much for your help,
>
> Guillermo
>
>
>
>
>
>
> On Wed, Jun 1, 2016 at 6:52 PM, Guillermo Fuentes
> <guillermo.fuentes at modernizingmedicine.com> wrote:
>> I'm now taking stack traces every minute and waiting for it to hang
>> again to check it. It happens usually under load but it's
>> unpredictable. Must likely tomorrow.
>> GUILLERMO FUENTES
>> SR. SYSTEMS ADMINISTRATOR
>>
>> 561-880-2998 x1337
>>
>> guillermo.fuentes at modmed.com
>>
>>
>>
>>
>>
>>
>> On Wed, Jun 1, 2016 at 2:03 PM, Rich Megginson <rmeggins at redhat.com> wrote:
>>> On 06/01/2016 10:37 AM, Guillermo Fuentes wrote:
>>>> Hi all,
>>>>
>>>> We are experiencing a similar issue like the one discussed in the
>>>> following thread but we are running FreeIPA 4.2 on CentOS 7.2:
>>>> https://www.redhat.com/archives/freeipa-users/2015-February/msg00205.html
>>>
>>> Are your stack traces similar?
>>>
>>>
>>>> LDAP service stops responding to queries (hangs). LDAP connections on
>>>> the server climb sometimes up to 10 times the normal amount and load
>>>> goes to 0. Then, the connections start to drop until they get to a
>>>> normal level and the LDAP service starts to respond to queries again.
>>>> This happens in between 3-5 minutes:
>>>>
>>>> Time,LDAP conn, Opened files(ns-slapd), File
>>>> Desc(ns-slapd),Threads(ns-slapd),Load1,Load5,Load15
>>>> 8:54:03,101,353,216,142,0.43,0.20,0.16
>>>> 8:55:02,108,359,221,142,0.19,0.18,0.15
>>>> 8:56:03,110,361,224,142,0.07,0.15,0.14
>>>> 8:57:14,117,383,246,142,0.15,0.16,0.15
>>>> 8:58:04,276,371,234,142,0.05,0.13,0.14
>>>> 8:59:05,469,371,234,142,0.02,0.11,0.13
>>>> 9:00:08,719,371,234,142,0.01,0.09,0.12
>>>> 9:01:18,1060,371,234,142,0.00,0.07,0.12
>>>> 9:02:10,742,371,233,142,0.10,0.09,0.12
>>>> 9:03:06,365,372,235,142,0.13,0.10,0.13
>>>> 9:04:04,262,379,242,142,0.87,0.29,0.19
>>>> 9:05:02,129,371,233,142,0.51,0.31,0.20
>>>> 9:06:03,126,377,240,142,0.42,0.33,0.22
>>>> 9:07:03,125,377,238,142,0.17,0.27,0.21
>>>>
>>>> Nothing is logged in the errors log file of the server having the
>>>> problem (ipa1 as an example).
>>>> In the replicas this is logged:
>>>> 8:59:05 -0400] NSMMReplicationPlugin - agmt="cn=meToipa1.example.com"
>>>> (ipa1:389): Unable to receive the response for a startReplication
>>>> extended operation to consumer (Timed out). Will retry later.
>>>> 9:01:05 -0400] NSMMReplicationPlugin - agmt="cn=meToipa1.example.com"
>>>> (ipa1:389): Unable to receive the response for a startReplication
>>>> extended operation to consumer (Timed out). Will retry later.
>>>>
>>>> Nothing is logged in the access log file until after ns-slapd starts
>>>> responding again:
>>>> ...
>>>> 8:57:00 -0400] conn=12384 fd=234 slot=234 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 8:57:00 -0400] conn=12385 fd=235 slot=235 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 8:57:00 -0400] conn=12386 fd=236 slot=236 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 8:57:00 -0400] conn=12387 fd=237 slot=237 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 8:57:00 -0400] conn=10384 op=1227 EXT oid="2.16.840.1.113730.3.5.12"
>>>> name="replication-multimaster-extop"
>>>> 8:57:00 -0400] conn=12324 op=8 RESULT err=0 tag=101 nentries=1 etime=0
>>>> 8:57:00 -0400] conn=8838 op=2545 EXT oid="2.16.840.1.113730.3.5.12"
>>>> name="replication-multimaster-extop"
>>>> 8:57:00 -0400] conn=8838 op=2545 RESULT err=0 tag=120 nentries=0 etime=0
>>>> 8:57:00 -0400] conn=10384 op=1227 RESULT err=0 tag=120 nentries=0 etime=0
>>>> 8:57:00 -0400] conn=12382 op=-1 fd=170 closed - B1
>>>> 8:57:00 -0400] conn=12383 op=0 SRCH base="" scope=0
>>>> filter="(objectClass=*)" attrs="supportedSASLMechanisms
>>>> defaultnamingcontext namingContexts schemanamingcontext saslrealm"
>>>> 8:57:00 -0400] conn=12384 op=-1 fd=234 closed - B1
>>>> 8:57:00 -0400] conn=12385 op=0 SRCH base="" scope=0
>>>> filter="(objectClass=*)" attrs="supportedSASLMechanisms
>>>> defaultnamingcontext namingContexts schemanamingcontext saslrealm"
>>>> 8:57:00 -0400] conn=12383 op=0 RESULT err=0 tag=101 nentries=1 etime=0
>>>> 8:57:00 -0400] conn=12386 op=-1 fd=236 closed - B1
>>>> 8:57:00 -0400] conn=12385 op=0 RESULT err=0 tag=101 nentries=1 etime=0
>>>> 8:57:00 -0400] conn=12387 op=0 SRCH base="" scope=0
>>>> filter="(objectClass=*)" attrs="supportedSASLMechanisms
>>>> defaultnamingcontext namingContexts schemanamingcontext saslrealm"
>>>> 8:57:00 -0400] conn=12387 op=0 RESULT err=0 tag=101 nentries=1 etime=0
>>>> 8:57:00 -0400] conn=12385 op=1 BIND dn="" method=sasl version=3
>>>> mech=GSSAPI
>>>> 8:57:00 -0400] conn=12387 op=1 BIND dn="" method=sasl version=3
>>>> mech=GSSAPI
>>>> 8:57:00 -0400] conn=12385 op=1 RESULT err=14 tag=97 nentries=0
>>>> etime=0, SASL bind in progress
>>>> 8:57:00 -0400] conn=12383 op=1 BIND dn="" method=sasl version=3
>>>> mech=GSSAPI
>>>> 8:57:00 -0400] conn=10384 op=1228 EXT oid="2.16.840.1.113730.3.5.5"
>>>> name="Netscape Replication End Session"
>>>> 8:57:00 -0400] conn=10384 op=1228 RESULT err=0 tag=120 nentries=0 etime=0
>>>> 8:57:00 -0400] conn=12383 op=1 RESULT err=14 tag=97 nentries=0
>>>> etime=0, SASL bind in progress
>>>> 9:02:00 -0400] conn=12388 fd=170 slot=170 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 9:02:00 -0400] conn=12389 fd=234 slot=234 SSL connection from
>>>> 172.20.0.24 to 172.20.2.45
>>>> 9:02:00 -0400] conn=12390 fd=236 slot=236 connection from local to
>>>> /var/run/slapd-EXAMPLE-COM.socket
>>>> 9:02:00 -0400] conn=12391 fd=238 slot=238 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 9:02:00 -0400] conn=12392 fd=239 slot=239 SSL connection from
>>>> 172.20.0.24 to 172.20.2.45
>>>> 9:02:00 -0400] conn=12393 fd=240 slot=240 connection from local to
>>>> /var/run/slapd-EXAMPLE-COM.socket
>>>> 9:02:00 -0400] conn=12394 fd=241 slot=241 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 9:02:00 -0400] conn=12395 fd=242 slot=242 SSL connection from
>>>> 172.20.0.24 to 172.20.2.45
>>>> 9:02:00 -0400] conn=12396 fd=243 slot=243 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 9:02:00 -0400] conn=12397 fd=244 slot=244 SSL connection from
>>>> 172.20.0.24 to 172.20.2.45
>>>> 9:02:00 -0400] conn=12398 fd=245 slot=245 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 9:02:00 -0400] conn=12400 fd=247 slot=247 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> 9:02:00 -0400] conn=12401 fd=248 slot=248 connection from 172.20.0.1
>>>> to 172.20.2.45
>>>> ...
>>>> 9:02:00 -0400] conn=12390 op=0 BIND dn="" method=sasl version=3
>>>> mech=GSSAPI
>>>> 9:02:00 -0400] conn=12388 op=-1 fd=170 closed - B1
>>>> 9:02:00 -0400] conn=12393 op=0 BIND dn="" method=sasl version=3
>>>> mech=GSSAPI
>>>> 9:02:00 -0400] conn=12391 op=0 SRCH base="" scope=0
>>>> filter="(objectClass=*)" attrs="supportedSASLMechanisms
>>>> defaultnamingcontext namingContexts schemanamingcontext saslrealm"
>>>> 9:02:00 -0400] conn=12394 op=-1 fd=241 closed - B1
>>>> 9:02:00 -0400] conn=12391 op=0 RESULT err=0 tag=101 nentries=1 etime=0
>>>> 9:02:00 -0400] conn=12396 op=0 SRCH base="" scope=0
>>>> filter="(objectClass=*)" attrs="supportedSASLMechanisms
>>>> defaultnamingcontext namingContexts schemanamingcontext saslrealm"
>>>> 9:02:00 -0400] conn=12396 op=0 RESULT err=0 tag=101 nentries=1 etime=0
>>>> 9:02:00 -0400] conn=12398 op=-1 fd=245 closed - B1
>>>> 9:02:00 -0400] conn=12400 op=0 SRCH base="" scope=0
>>>> filter="(objectClass=*)" attrs="supportedSASLMechanisms
>>>> defaultnamingcontext namingContexts schemanamingcontext saslrealm"
>>>> 9:02:00 -0400] conn=12400 op=0 RESULT err=0 tag=101 nentries=1 etime=0
>>>> 9:02:00 -0400] conn=12401 op=-1 fd=248 closed - B1
>>>> 9:02:00 -0400] conn=12391 op=1 ABANDON targetop=NOTFOUND msgid=1
>>>> 9:02:00 -0400] conn=12396 op=1 ABANDON targetop=NOTFOUND msgid=1
>>>> 9:02:00 -0400] conn=12400 op=1 ABANDON targetop=NOTFOUND msgid=1
>>>> 9:02:00 -0400] conn=12391 op=2 UNBIND
>>>> 9:02:00 -0400] conn=12396 op=2 UNBIND
>>>> 9:02:00 -0400] conn=12391 op=2 fd=238 closed - U1
>>>> 9:02:00 -0400] conn=12396 op=2 fd=243 closed - U1
>>>> 9:02:00 -0400] conn=12400 op=2 UNBIND
>>>> 9:02:00 -0400] conn=12400 op=2 fd=247 closed - U1
>>>> ...
>>>>
>>>>
>>>> Environment:
>>>> # cat /etc/redhat-release
>>>> CentOS Linux release 7.2.1511 (Core)
>>>>
>>>> # rpm -qa ipa*
>>>> ipa-server-4.2.0-15.0.1.el7.centos.6.1.x86_64
>>>> ipa-python-4.2.0-15.0.1.el7.centos.6.1.x86_64
>>>> ipa-admintools-4.2.0-15.0.1.el7.centos.6.1.x86_64
>>>> ipa-server-trust-ad-4.2.0-15.0.1.el7.centos.6.1.x86_64
>>>> ipa-client-4.2.0-15.0.1.el7.centos.6.1.x86_64
>>>> ipa-server-dns-4.2.0-15.0.1.el7.centos.6.1.x86_64
>>>>
>>>> # rpm -qa 389*
>>>> 389-ds-base-libs-1.3.4.0-30.el7_2.x86_64
>>>> 389-ds-base-1.3.4.0-30.el7_2.x86_64
>>>>
>>>> We have 4 FreeIPA servers with replication working fine between them.
>>>> ipa1 is handling LDAP authentication for +400 clients and has been
>>>> tunned as recommended per
>>>>
>>>> https://access.redhat.com/documentation/en-US/Red_Hat_Directory_Server/8.2/html/Performance_Tuning_Guide/system-tuning.html
>>>>
>>>> Is this a known issue?
>>>> Any idea what can be causing ns-slapd to hang?
>>>>
>>>> Thanks in advance!
>>>>
>>>> Guillermo
>>>>
>>> --
>>> Manage your subscription for the Freeipa-users mailing list:
>>> https://www.redhat.com/mailman/listinfo/freeipa-users
>>> Go to http://freeipa.org for more info on the project