[
https://issues.apache.org/jira/browse/SCM-300?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17960961#comment-17960961
]
ASF GitHub Bot commented on SCM-300:
------------------------------------
jira-importer opened a new issue, #519:
URL: https://github.com/apache/maven-scm/issues/519
**[Hilco
Wijbenga](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=hilco)**
opened
**[SCM-300](https://issues.apache.org/jira/browse/SCM-300?redirect=false)** and
commented
We've been trying to set up Continuum and consequently we started
running some builds. It all worked nicely until the CVS server ran out
of inodes in its /tmp directory. It looks like it's not the normal
checkouts that people do that's causing this ... it appears to be
Continuum.
We tried a small build with Continuum and lo and behold: /tmp had a
new directory that was not deleted at the end of the checkout. So
we're fairly certain it's something Continuum related.
CVS creates a sort of cache directory (in /tmp) for each checkout (it
stores the CVS directories, not any of the files that are being
checked out) and this directory (and all of its **many** subdirectories)
are not being deleted. Apparently, it doesn't notice that the checkout
has finished. Something like a session not being closed. After three
days of this our server runs out of inodes and nothing works anymore.
We've created a cron job to remove these CVS directories from /tmp at
the end of the day so we have a workaround.
---
**Affects:** 1.0-rc1
> CVS doesn't close it's session after checkout
> ---------------------------------------------
>
> Key: SCM-300
> URL: https://issues.apache.org/jira/browse/SCM-300
> Project: Maven SCM (Moved to GitHub Issues)
> Issue Type: Bug
> Components: maven-scm-provider-cvs
> Affects Versions: 1.0-rc1
> Environment: Linux, Sun JDK 1.4. Continuum built from trunk.
> Reporter: Hilco Wijbenga
> Assignee: Emmanuel Venisse
> Priority: Major
> Fix For: 1.0
>
>
> We've been trying to set up Continuum and consequently we started
> running some builds. It all worked nicely until the CVS server ran out
> of inodes in its /tmp directory. It looks like it's not the normal
> checkouts that people do that's causing this ... it appears to be
> Continuum.
> We tried a small build with Continuum and lo and behold: /tmp had a
> new directory that was not deleted at the end of the checkout. So
> we're fairly certain it's something Continuum related.
> CVS creates a sort of cache directory (in /tmp) for each checkout (it
> stores the CVS directories, not any of the files that are being
> checked out) and this directory (and all of its *many* subdirectories)
> are not being deleted. Apparently, it doesn't notice that the checkout
> has finished. Something like a session not being closed. After three
> days of this our server runs out of inodes and nothing works anymore.
> We've created a cron job to remove these CVS directories from /tmp at
> the end of the day so we have a workaround.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
