FYI: We're doing something similar in the GHA workflow for Apache Flink [1].
[1] https://github.com/apache/flink/blob/master/.github/actions/job_init/action.yml#L54-L69 On Wed, Jul 10, 2024 at 3:53 PM João Boto <eskabe...@apache.org> wrote: > I will send this with better format.. > Sorry for that > > On 2024/07/10 13:46:43 Joao Boto wrote: > > Hi, > > > > During the refactor related to FLIP-449[1], we start seeing this error on > > CI like: > > > > > - "You are running out of disk space. The runner will stop working when > > > the machine runs out of disk space. Free space left: 0 MB" > > > > - "Warning: ForkStarter IOException: Unable to create file for report: > No > > > space left on device. See the dump file > > > > /home/runner/work/flink-connector-jdbc/flink-connector-jdbc/flink-connector-jdbc-oceanbase/target/surefire-reports/2024-07-09T09-59-05_229-jvmRun4.dumpstream" > > > > > > The runner has 21gb free at start, and the docker images used have 12gb.. > > (this could grow if we add more databases) > > Free space > > > > > Filesystem Size Used Avail Use% Mounted on > > > 12 > > > < > https://github.com/apache/flink-connector-jdbc/actions/runs/9854907586/job/27208719095#step:2:13 > >/dev/root > > > 73G 52G 21G 72% / > > > > > > > We could consider that the maven could use another 3gb (in jar, class > files > > etc), but we should have another 5gb of free space.. > > > > I create a copy of the workflow to allow add some steps, making some > > cleanup in the runner to get more space, and now I can run the tests > > > > > Filesystem Size Used Avail Use% Mounted on > > > 12 > > > < > https://github.com/apache/flink-connector-jdbc/actions/runs/9854907586/job/27208719095#step:4:13 > >/dev/root > > > 73G 34G 40G 46% / > > > > > > > these are the changes: > > > https://github.com/eskabetxe/flink-connector-jdbc/blob/FLINK-35363/.github/workflows/clean_space.yml#L47-L54 > > > > Could someone help us here? > > Should we add to > > "apache/flink-connector-shared-utils/.github/workflows/ci.yml@ci_utils" > an > > option to clean the runner? > > Another option could be change docker to store files on "/mnt" that have > > 66GB free (dont find any way to do this) > > > > > /dev/sdb1 74G 4.1G 66G 6% /mnt > > > > > > > > > this is the last run: > > https://github.com/apache/flink-connector-jdbc/actions/runs/9854907626 > > this is the last run of clean CI: > > https://github.com/apache/flink-connector-jdbc/actions/runs/9854907586 > > > > > > > > Hi Team, > > > > During the refactor related to FLIP-449, we have started encountering the > > following errors on CI: > > > > > "You are running out of disk space. The runner will stop working when > the > > > machine runs out of disk space. Free space left: 0 MB." > > > > > > > > > "Warning: ForkStarter IOException: Unable to create file for report: No > > > space left on device. See the dump file > > > > /home/runner/work/flink-connector-jdbc/flink-connector-jdbc/flink-connector-jdbc-oceanbase/target/surefire-reports/2024-07-09T09-59-05_229-jvmRun4.dumpstream." > > > > > > Initially, the runner has 21GB free space, and the docker images used > take > > up 12GB. This usage could increase if more databases are added. The free > > space details are as follows: > > > > Filesystem Size Used Avail Use% Mounted on > > > /dev/root 73G 52G 21G 72% / > > > > Considering Maven might use another 3GB (for jars, class files, etc.), we > > should still have around 5GB of free space. > > > > I have created a copy of the workflow and added some steps to clean up > the > > runner to free up more space. With these changes, I am now able to run > the > > tests successfully. The updated free space details are: > > > > Filesystem Size Used Avail Use% Mounted on > > > /dev/root 73G 34G 40G 46% / > > > > You can find the changes I made here: > > > https://github.com/eskabetxe/flink-connector-jdbc/blob/FLINK-35363/.github/workflows/clean_space.yml#L47-L54 > > > > Could someone assist us in resolving this issue? > > > > Should we add an option to clean the runner in > > apache/flink-connector-shared-utils/.github/workflows/ci.yml@ci_utils? > > > > Another option might be to configure Docker to store files on /mnt, > which > > has 66GB free space: > > > > Filesystem Size Used Avail Use% Mounted on > > > /dev/sdb1 74G 4.1G 66G 6% /mnt > > > > However, I haven't found a way to do this. > > > > Here are the details of the last runs: > > > > - Last run of CI: > > > https://github.com/apache/flink-connector-jdbc/actions/runs/9854907626 > > - Last run of clean CI: > > > https://github.com/apache/flink-connector-jdbc/actions/runs/9854907586 > > > > Thank you for your help! > > > > Best regards, > > > > [1] > > > https://cwiki.apache.org/confluence/display/FLINK/FLIP-449%3A+Reorganization+of+flink-connector-jdbc > > >
