[Pulp-list] Celery memory usage during RPM repo copy

Richard Gray richard.gray at smxemail.com
Thu Feb 16 21:07:00 UTC 2017


I have a local RPM repository that syncs from an internet mirror of the 
CentOS 7.3 updates repository. When I try to create a local copy of this 
mirror, the celery processes balloon, using all available memory until 
the job eventually fails as follows:

$ pulp-admin rpm repo copy all --from-repo-id 
centos-7-updates-x86_64-live --to-repo-id 
This command may be exited via ctrl+c without affecting the request.

An internal error occurred on the Pulp server:

RequestException: GET request on
/pulp/api/v2/tasks/776894cc-b524-4f2f-bd30-c8c4d8237c31/ failed with 500 -
[Errno 12] Cannot allocate memory

As often as not, this renders my Pulp VM inaccessible, requiring a hard 
reset in order to access it again. This is Pulp 2.12.0 on CentOS 
7.2.1511. I presume this is the same issue described in 
https://pulp.plan.io/issues/1779, but given 10 months since that ticket 
was updated, I figured I'd ask here.

I've tried setting PULP_MAX_TASKS_PER_CHILD=2 in 
/etc/default/pulp_workers, but it doesn't seem to make any difference.

Is there a fix or workaround for this problem, or even a means to 
estimate how much memory a server needs to be able to copy a repo of a 
given size? This VM has 8GB, and I could add more, but it seems faintly 
ridiculous that it should be required.


*Richard Gray* | Operations Technical Lead
*DDI:* +64 9 950 2196 *Fax:* +64 9 302 0518
*Mobile:* +64 21 050 8178 *Freephone:*0800 SMX SMX (769 769)
*SMX Limited:* Level 15, 19 Victoria Street West, Auckland, New Zealand
*Web:* http://smxemail.com
SMX | Cloud Email Hosting & Security


This email has been filtered by SMX. For more info visit http://smxemail.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/pulp-list/attachments/20170217/fabc39a7/attachment.htm>

More information about the Pulp-list mailing list