Skip to content

Increase event page size to 1000#6937

Merged
mdubya66 merged 1 commit intovmware:masterfrom
cgtexmex:eventful
Dec 10, 2017
Merged

Increase event page size to 1000#6937
mdubya66 merged 1 commit intovmware:masterfrom
cgtexmex:eventful

Conversation

@cgtexmex
Copy link
Contributor

@cgtexmex cgtexmex commented Dec 8, 2017

Increasing the event page size on the portlayer
to help with missing events.

Helps with #6384, #6851

Increasing the event page size on the portlayer
to help with missing events.

Helps with vmware#6384, vmware#6851
@hickeng
Copy link
Contributor

hickeng commented Dec 8, 2017

Could you run test Group0-Bugs/5343 against this to confirm sane recovery when the session is dropped.
My require manual inspection of logs to determine 'sane'.

@cgtexmex
Copy link
Contributor Author

cgtexmex commented Dec 8, 2017

@hickeng ran that test locally and am concerned by the amount of event spew on VCH startup -- the event collector is created and the last 1K events on that host / cluster before it's powerOn will be seen by the VCH. That could be helpful in avoiding missed events in the case of an outage, but unhelpful when it's a new VCH. In the case of a new VCH we'll log and callback for many events that have nothing to do with the VCH. It added about 70KB to the portLayer log...

Copy link
Contributor

@mhagen-vmware mhagen-vmware left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Lgtm

@mdubya66 mdubya66 merged commit 7796336 into vmware:master Dec 10, 2017
cgtexmex pushed a commit to cgtexmex/vic that referenced this pull request Dec 13, 2017
Increasing the event page size on the portlayer
to help with missing events.
Helps with vmware#6384, vmware#6851
mhagen-vmware pushed a commit that referenced this pull request Dec 13, 2017
Increasing the event page size on the portlayer
to help with missing events.
Helps with #6384, #6851
mdharamadas1 added a commit to mdharamadas1/vic that referenced this pull request Dec 15, 2017
hickeng added a commit to hickeng/vic that referenced this pull request Dec 21, 2017
Speculative that there are two issues:
1. leaking VCHs from group23
2. event page size

Hypothesis is that either of those can result in builds overwhelming hostd
and the combination has made addressing one in isolation return false
negatives for impact.

This reverts commit 7796336.
hickeng pushed a commit to hickeng/vic that referenced this pull request Dec 21, 2017
Increasing the event page size on the portlayer
to help with missing events.
Helps with vmware#6384, vmware#6851
hickeng added a commit to hickeng/vic that referenced this pull request Dec 21, 2017
Speculative that there are two issues:
1. leaking VCHs from group23
2. event page size

Hypothesis is that either of those can result in builds overwhelming hostd
and the combination has made addressing one in isolation return false
negatives for impact.

This reverts commit 7796336.
hickeng added a commit to hickeng/vic that referenced this pull request Dec 21, 2017
This reverts commit 7796336 because it's
apparent that increasing the page size to this extent can cause hostd to
both hit its resource limits and to drastically fragment its heap.
hickeng added a commit that referenced this pull request Dec 21, 2017
* Fix leak of VCHs after test runs

The delete tests for vic-machine and vic-machine-service leaks VCHs.
For the service it's because the tests deploy VCHs directly that are not
cleaned up.
For vic-machine base it's because we render the VCH invalid by moving the
endpointVM in such a manner that the deletion fails without explicit
cleanup after.

* Revert "Increase event page size to 1000 (#6937)"

This reverts commit 7796336 because it's
apparent that increasing the page size to this extent can cause hostd to
both hit its resource limits and drastically fragment its heap.
hickeng added a commit to hickeng/vic that referenced this pull request Dec 21, 2017
* Fix leak of VCHs after test runs

The delete tests for vic-machine and vic-machine-service leaks VCHs.
For the service it's because the tests deploy VCHs directly that are not
cleaned up.
For vic-machine base it's because we render the VCH invalid by moving the
endpointVM in such a manner that the deletion fails without explicit
cleanup after.

* Revert "Increase event page size to 1000 (vmware#6937)"

This reverts commit 7796336 because it's
apparent that increasing the page size to this extent can cause hostd to
both hit its resource limits and drastically fragment its heap.
dougm added a commit that referenced this pull request Dec 21, 2017
An alternative to increasing the collector page size.  It will reduce the throughput to the event collector and hence reduce event misses.

See issues #6937 and #6998
hickeng added a commit that referenced this pull request Dec 21, 2017
* Fix leak of VCHs after test runs

The delete tests for vic-machine and vic-machine-service leaks VCHs.
For the service it's because the tests deploy VCHs directly that are not
cleaned up.
For vic-machine base it's because we render the VCH invalid by moving the
endpointVM in such a manner that the deletion fails without explicit
cleanup after.

* Revert "Increase event page size to 1000 (#6937)"

This reverts commit 7796336 because it's
apparent that increasing the page size to this extent can cause hostd to
both hit its resource limits and drastically fragment its heap.
dougm added a commit that referenced this pull request Dec 21, 2017
An alternative to increasing the collector page size.  It will reduce the throughput to the event collector and hence reduce event misses.

See issues #6937 and #6998
hickeng pushed a commit to hickeng/vic that referenced this pull request Dec 29, 2017
An alternative to increasing the collector page size.  It will reduce the throughput to the event collector and hence reduce event misses.

See issues vmware#6937 and vmware#6998
hickeng pushed a commit to hickeng/vic that referenced this pull request Dec 29, 2017
An alternative to increasing the collector page size.  It will reduce the throughput to the event collector and hence reduce event misses.

See issues vmware#6937 and vmware#6998
hickeng pushed a commit that referenced this pull request Dec 29, 2017
An alternative to increasing the collector page size.  It will reduce the throughput to the event collector and hence reduce event misses.

See issues #6937 and #6998
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants