Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

regression.sh: use a zram device for /d when available #213

Open
wants to merge 2 commits into
base: master
Choose a base branch
from

Conversation

xhernandez
Copy link
Contributor

It has been observed that AWS builder instances are very slow sometimes, apparently because of a limited IOPS. To avoid this issue, the script regression.sh has been modified to try to mount /d on a zram device, which will drastically reduce the number of IOPS sent to AWS disks. If the zram device cannot be created for some reason, it falls back to the old loop device.

This change should avoid spurious timeout failures caused by too long executions, and it should also prevent failures in tests that are time-dependent.

It has been observed that AWS builder instances are very slow sometimes,
apparently because of a limited IOPS. To avoid this issue, the script
regression.sh has been modified to try to mount /d on a zram device,
which will drastically reduce the number of IOPS sent to AWS disks. If
the zram device cannot be created for some reason, it falls back to the
old loop device.

This change should avoid spurious timeout failures caused by too long
executions, and it should also prevent failures in tests that are
time-dependent.

Signed-off-by: Xavi Hernandez <[email protected]>
@xhernandez
Copy link
Contributor Author

@mscherer this change should help reduce spurious failures in the regression suite. Could you take a look, please ?

If I'm nor wrong, /d is also used to store the built programs. This should be changed. Otherwise all compiled programs will be deleted when regression.sh is run.

regression.sh Outdated Show resolved Hide resolved
@mykaul
Copy link

mykaul commented May 8, 2023

(unrelated to this patch, we should also build against zram device instead of the EBS disks)

@xhernandez
Copy link
Contributor Author

(unrelated to this patch, we should also build against zram device instead of the EBS disks)

I tried this and it seems to work with the current tests, but builder machines are very limited (2 cores, 4 GiB if I'm not wrong). So adding more pressure to memory may be bad if some test really requires to use a significant amount of space.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants