fix(cluster_aws): moved var-lib-scylla.mount away after restart #2839
Conversation
Since scylla_create_devices fails if var-lib-scylla.mount already exists during its execution, I moved var-lib-scylla.mount from its regular location
| @@ -675,6 +675,12 @@ def restart(self): | |||
| try: | |||
| self.stop_scylla_server(verify_down=False) | |||
|
|
|||
| # Moving var-lib-scylla.mount away, since scylla_create_devices fails if it already exists | |||
There was a problem hiding this comment.
var-lib-scylla.mount file is created by scylla_setup.
what's the error if var-lib-scylla.mount exists?
If the var-lib-scylla.mount is moved, will the RAID be mounted correctly after reboot?
There was a problem hiding this comment.
We don’t know, isn’t it created by create_devices?
We saw that create devices is failing when this file exist
There was a problem hiding this comment.
seem like @syuu1228 did this change few months ago:
scylladb/scylladb@c0b2933
I think we should remove it file inside create_devices script, if that's the case
There was a problem hiding this comment.
We don’t know, isn’t it created by create_devices?
We saw that create devices is failing when this file exist
In restart() we call scylla_create_devices script to setup RAID, the existing var-lib-scylla.mount won't effect it.
it only effects scylla_setup script, but we didn't call it in restart().
There was a problem hiding this comment.
seem like @syuu1228 did this change few months ago:
scylladb/scylla@c0b2933
The commit changed scylla_setup, but we won't call it in restart().
I think we should remove it file inside create_devices script, if that's the case
@ShlomiBalalis Need to check the source of the error someone touched, is it from scylla_setup, why it's called.
Currently we only directly call scylla_setup script for GCE backend in init stage.
A flag file /etc/scylla/machine_image_configured will be created after scylla_setup -> scylla_image_setup succeeds, then scylla-image-setup won't be called again.
|
There are two errors in executing scylla_setup or scylla_raid_setup:
I guess @ShlomiBalalis touched Error 2, I'm ok with this PR. |
| @@ -675,6 +675,12 @@ def restart(self): | |||
| try: | |||
| self.stop_scylla_server(verify_down=False) | |||
|
|
|||
| # Moving var-lib-scylla.mount away, since scylla_create_devices fails if it already exists | |||
| mount_path = "/etc/systemd/system/var-lib-scylla.mount" | |||
| existance_check = self.remoter.run(f'sudo test -e {mount_path}', ignore_status=True) | |||
There was a problem hiding this comment.
It's fine to directly remove it without checking if it exists or not.
sudo rm -f /etc/systemd/system/var-lib-scylla.mount
There was a problem hiding this comment.
@roy wanted me to save the file for debugging purposes
|
@ShlomiBalalis I'm merging this, but please open a bug, so we'll fix it in the correct place |
|
@fruch Why do we need a bug for it in scylla-machine-image? |
We need to handle this inside |
Since scylla_create_devices fails if var-lib-scylla.mount already exists during its execution,
I moved var-lib-scylla.mount from its regular location
PR pre-checks (self review)
backportlabelssdcm/sct_config.py)unit-test/folder)