[JIRA] (OVIRT-1161) Nightly publisher is broken
by Barak Korren (oVirt JIRA)
[ https://ovirt-jira.atlassian.net/browse/OVIRT-1161?page=com.atlassian.jir... ]
Barak Korren reassigned OVIRT-1161:
-----------------------------------
Assignee: Barak Korren (was: infra)
> Nightly publisher is broken
> ---------------------------
>
> Key: OVIRT-1161
> URL: https://ovirt-jira.atlassian.net/browse/OVIRT-1161
> Project: oVirt - virtualization made easy
> Issue Type: Bug
> Components: Repositories Mgmt
> Reporter: Barak Korren
> Assignee: Barak Korren
> Priority: Highest
>
> Nightly publisher is currently broken, this has caused some of the oVirt snapshot repos to become empty, which in turn caused OST failures, among others.
> The error output from the nightly publisher job is as follows:
> {code}
> [/usr/local/libexec/releng-tools/scan_for_artifacts.sh] ##### Starting artifacts publishing Sun Feb 19 09:00:01 UTC 2017
> FATAL: We still have files under /home/jenkins/artifacts/ovirt-master-snapshot.ready. Please recheck
> FATAL: Failed inside the loop
> {code}
> This is probably caused by YUM metadata that has been placed in the {{/home/jenkins/artifacts/ovirt-master-snapshot.ready}} by repoman that had been added in OVIRT-1152.
> This as also surfaced another issue - if the {{\*-snapshot.ready}} directory exists while the nightly publisher job runs (which means the cron job did not finish yet), it will happily move the {{\*-snaphot.tmp}} directory into it. What it should do instead is wait a while for it to vanish, and then fail if it does not.
--
This message was sent by Atlassian JIRA
(v1000.769.1#100032)
7 years, 9 months
[JIRA] (OVIRT-1161) Nightly publisher is broken
by Barak Korren (oVirt JIRA)
[ https://ovirt-jira.atlassian.net/browse/OVIRT-1161?page=com.atlassian.jir... ]
Barak Korren updated OVIRT-1161:
--------------------------------
Description:
Nightly publisher is currently broken, this has caused some of the oVirt snapshot repos to become empty, which in turn caused OST failures, among others.
The error output from the nightly publisher job is as follows:
{code}
[/usr/local/libexec/releng-tools/scan_for_artifacts.sh] ##### Starting artifacts publishing Sun Feb 19 09:00:01 UTC 2017
FATAL: We still have files under /home/jenkins/artifacts/ovirt-master-snapshot.ready. Please recheck
FATAL: Failed inside the loop
{code}
This is probably caused by YUM metadata that has been placed in the {{/home/jenkins/artifacts/ovirt-master-snapshot.ready}} by repoman that had been added in OVIRT-1152.
This as also surfaced another issue - if the {{\*-snapshot.ready}} directory exists while the nightly publisher job runs (which means the cron job did not finish yet), it will happily move the {{\*-snaphot.tmp}} directory into it. What it should do instead is wait a while for it to vanish, and then fail if it does not.
was:
Nightly publisher is currently broken, this has caused some of the oVirt snapshot repos to become empty, which in turn caused OST failures, among others.
The error output from the nightly publisher job is as follows:
{code}
[/usr/local/libexec/releng-tools/scan_for_artifacts.sh] ##### Starting artifacts publishing Sun Feb 19 09:00:01 UTC 2017
FATAL: We still have files under /home/jenkins/artifacts/ovirt-master-snapshot.ready. Please recheck
FATAL: Failed inside the loop
{code}
This is probably caused by YUM metadata that has been placed in the {{/home/jenkins/artifacts/ovirt-master-snapshot.ready}} by repoman that had been added in OVIRT-1152.
This as also surfaced another issue - if the {{*-snapshot.ready}} directory exists while the nightly publisher job runs (which means the cron job did not finish yet), it will happily move the {{*-snaphot.tmp}} directory into it. What it should do instead is wait a while for it to vanish, and then fail if it does not.
> Nightly publisher is broken
> ---------------------------
>
> Key: OVIRT-1161
> URL: https://ovirt-jira.atlassian.net/browse/OVIRT-1161
> Project: oVirt - virtualization made easy
> Issue Type: Bug
> Components: Repositories Mgmt
> Reporter: Barak Korren
> Assignee: infra
> Priority: Highest
>
> Nightly publisher is currently broken, this has caused some of the oVirt snapshot repos to become empty, which in turn caused OST failures, among others.
> The error output from the nightly publisher job is as follows:
> {code}
> [/usr/local/libexec/releng-tools/scan_for_artifacts.sh] ##### Starting artifacts publishing Sun Feb 19 09:00:01 UTC 2017
> FATAL: We still have files under /home/jenkins/artifacts/ovirt-master-snapshot.ready. Please recheck
> FATAL: Failed inside the loop
> {code}
> This is probably caused by YUM metadata that has been placed in the {{/home/jenkins/artifacts/ovirt-master-snapshot.ready}} by repoman that had been added in OVIRT-1152.
> This as also surfaced another issue - if the {{\*-snapshot.ready}} directory exists while the nightly publisher job runs (which means the cron job did not finish yet), it will happily move the {{\*-snaphot.tmp}} directory into it. What it should do instead is wait a while for it to vanish, and then fail if it does not.
--
This message was sent by Atlassian JIRA
(v1000.769.1#100032)
7 years, 9 months
[JIRA] (OVIRT-1161) Nightly publisher is broken
by Barak Korren (oVirt JIRA)
[ https://ovirt-jira.atlassian.net/browse/OVIRT-1161?page=com.atlassian.jir... ]
Barak Korren updated OVIRT-1161:
--------------------------------
Epic Link: OVIRT-403
> Nightly publisher is broken
> ---------------------------
>
> Key: OVIRT-1161
> URL: https://ovirt-jira.atlassian.net/browse/OVIRT-1161
> Project: oVirt - virtualization made easy
> Issue Type: Bug
> Components: Repositories Mgmt
> Reporter: Barak Korren
> Assignee: infra
> Priority: Highest
>
> Nightly publisher is currently broken, this has caused some of the oVirt snapshot repos to become empty, which in turn caused OST failures, among others.
> The error output from the nightly publisher job is as follows:
> {code}
> [/usr/local/libexec/releng-tools/scan_for_artifacts.sh] ##### Starting artifacts publishing Sun Feb 19 09:00:01 UTC 2017
> FATAL: We still have files under /home/jenkins/artifacts/ovirt-master-snapshot.ready. Please recheck
> FATAL: Failed inside the loop
> {code}
> This is probably caused by YUM metadata that has been placed in the {{/home/jenkins/artifacts/ovirt-master-snapshot.ready}} by repoman that had been added in OVIRT-1152.
> This as also surfaced another issue - if the {{*-snapshot.ready}} directory exists while the nightly publisher job runs (which means the cron job did not finish yet), it will happily move the {{*-snaphot.tmp}} directory into it. What it should do instead is wait a while for it to vanish, and then fail if it does not.
--
This message was sent by Atlassian JIRA
(v1000.769.1#100032)
7 years, 9 months
[JIRA] (OVIRT-1161) Nightly publisher is broken
by Barak Korren (oVirt JIRA)
Barak Korren created OVIRT-1161:
-----------------------------------
Summary: Nightly publisher is broken
Key: OVIRT-1161
URL: https://ovirt-jira.atlassian.net/browse/OVIRT-1161
Project: oVirt - virtualization made easy
Issue Type: Bug
Components: Repositories Mgmt
Reporter: Barak Korren
Assignee: infra
Priority: Highest
Nightly publisher is currently broken, this has caused some of the oVirt snapshot repos to become empty, which in turn caused OST failures, among others.
The error output from the nightly publisher job is as follows:
{code}
[/usr/local/libexec/releng-tools/scan_for_artifacts.sh] ##### Starting artifacts publishing Sun Feb 19 09:00:01 UTC 2017
FATAL: We still have files under /home/jenkins/artifacts/ovirt-master-snapshot.ready. Please recheck
FATAL: Failed inside the loop
{code}
This is probably caused by YUM metadata that has been placed in the {{/home/jenkins/artifacts/ovirt-master-snapshot.ready}} by repoman that had been added in OVIRT-1152.
This as also surfaced another issue - if the {{*-snapshot.ready}} directory exists while the nightly publisher job runs (which means the cron job did not finish yet), it will happily move the {{*-snaphot.tmp}} directory into it. What it should do instead is wait a while for it to vanish, and then fail if it does not.
--
This message was sent by Atlassian JIRA
(v1000.769.1#100032)
7 years, 9 months
[JIRA] (OVIRT-1154) Make jenkins-whitelist use Gerrit groups
by Barak Korren (oVirt JIRA)
[ https://ovirt-jira.atlassian.net/browse/OVIRT-1154?page=com.atlassian.jir... ]
Barak Korren commented on OVIRT-1154:
-------------------------------------
We don't want to significantly slow down the CI because of this, a good guideline should be for the whole whitelist matching process to take:
# At most 1 network fetch (right not if fetches once via cloning from Gerrit)
# At most O\(n) time to search the match, where 'n' is the *number of contributors* as opposed to *number of patches*
AFAIK we don't keep lists of contributors anywhere ATM, so we'll probably need to scan patches to see who contributed. We may not be able to practically do this fast enough. We will need to check if Git or Gerrit can do this efficiently for us ahead of time.
We also need some proper definition of who is considered a contributor. I don't think that someone with a merged patch ever is a good heuristic, should probably be something like N patches merged in the last M months.
> Make jenkins-whitelist use Gerrit groups
> ----------------------------------------
>
> Key: OVIRT-1154
> URL: https://ovirt-jira.atlassian.net/browse/OVIRT-1154
> Project: oVirt - virtualization made easy
> Issue Type: Improvement
> Components: Jenkins
> Reporter: Barak Korren
> Assignee: infra
>
> Right now our white list is based on a text file with email addresses in it. Its kinda silly when we already have lists and groups of users in Gerrit.
> Some suggestions:
> * make any members of a project's "maintainers" group automatically white-listed
> * make a specific group for other white-listed people
--
This message was sent by Atlassian JIRA
(v1000.769.1#100032)
7 years, 9 months