brocade cli _ssh_execute does not scale for concurrent deployments

Bug #1436594 reported by Gerald McBrearty
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Cinder
Fix Released
Medium
Pradeep Sathasivam

Bug Description

In cinder/zonemanager/drivers/brocade/brcd_fc_zone_client_cli.py _ssh_execute(),
there is a random sleep of 0.2 to 5.0 seconds between the time the command is sent and when it sends ZoneConstant.
YES to confirm the operation.

We are seeing significant delays in VM deployment that we have attributed to this sleep with concurrent
deploys of 10 or more VMs.

With this sleep removed we have not seen other errors introduced by immediately sending YES.

Mike Perez (thingee)
tags: added: brocade fibrechannel zonemanager
Angela Smith (aallen-m)
Changed in cinder:
assignee: nobody → Angela Smith (aallen-m)
Revision history for this message
Gerald McBrearty (gfm-r) wrote :

Angela would you like me to contribute this minor fix?

Revision history for this message
Angela Smith (aallen-m) wrote :

Thanks Gerard. Our team is currently testing removal of the sleep to be positive it does not cause any adverse side effects. We will submit for review once we are certain the sleep is not needed.

Changed in cinder:
status: New → In Progress
Changed in cinder:
assignee: Angela Smith (aallen-m) → Pradeep Sathasivam (spradeepv)
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to cinder (master)

Fix proposed to branch: master
Review: https://review.openstack.org/171505

Jay Bryant (jsbryant)
Changed in cinder:
importance: Undecided → Medium
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to cinder (master)

Reviewed: https://review.openstack.org/171505
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=e2980d164965f2e2cae31452774d11e2552fffe2
Submitter: Jenkins
Branch: master

commit e2980d164965f2e2cae31452774d11e2552fffe2
Author: Pradeep Sathasivam <email address hidden>
Date: Wed Apr 8 12:56:23 2015 +0530

    Removed sleep before 'YES' is sent to confim an operation

    Removed sleep between the time a command is sent
    and 'YES' is sent to confim the operation. We have
    discussed this internally and have concluded that
    the sleep was mistakenly added in the initial coding
    from an example of the SSH code. We have done full
    functionality testing of the zoning operations on
    Brocade switches with the sleep removed and have
    found no side effects from the removal

    Additionally, no other Brocade management
    applications which perform zoning on our switches
    require sleep intervals between execution of
    commands. We are confident that the sleep is not
    necessary and can be removed

    Change-Id: I2ac6e50ff3d231b4d404f1a0be29f907e8d20458
    Closes-Bug: #1436594

Changed in cinder:
status: In Progress → Fix Committed
Revision history for this message
Yucong Feng (yfeng) wrote :

Can this fix be extended to the Cisco CLI as well? Or would there be need of another launchpad defect?

/zonemanager/drivers/cisco/cisco_fc_zone_client_cli.py _ssh_execute() ?

Revision history for this message
Angela Smith (aallen-m) wrote :

Please file a new defect. Thanks!

Thierry Carrez (ttx)
Changed in cinder:
milestone: none → liberty-1
status: Fix Committed → Fix Released
Thierry Carrez (ttx)
Changed in cinder:
milestone: liberty-1 → 7.0.0
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.