Support disabled mode in frostfs-cli control shards set-mode #917

Closed
opened 2024-01-17 08:44:05 +00:00 by BrookDaniel · 4 comments

EDIT by @fyrchik :
Changing the config has some problems: in particular not being able to easily wait for completion.
control service command will solve this problem.

EDIT by @fyrchik : Changing the config has some problems: in particular not being able to easily wait for completion. `control` service command will solve this problem.
BrookDaniel added the
triage
label 2024-01-17 08:44:05 +00:00
fyrchik added the
frostfs-node
label 2024-01-17 08:44:51 +00:00
fyrchik added this to the v0.39.0 milestone 2024-01-17 08:45:05 +00:00
fyrchik added
enhancement
good first issue
and removed
triage
labels 2024-01-17 08:45:42 +00:00

While we are here, let's also take care about shard-modes.md document.
We should mention possible problems with this appoach: configuration represents the desireable state, thus any restart/SIGHUP will lose control service changes.

While we are here, let's also take care about shard-modes.md document. We should mention possible problems with this appoach: configuration represents the desireable state, thus any restart/SIGHUP will lose control service changes.
fyrchik modified the milestone from v0.39.0 to v0.38.0 2024-01-17 08:46:58 +00:00
Collaborator

We should mention possible problems with this appoach: configuration represents the desireable state, thus any restart/SIGHUP will lose control service changes.

We can disable a shard until it gets really disabled, not caring about service changes reset?

> We should mention possible problems with this appoach: configuration represents the desireable state, thus any restart/SIGHUP will lose control service changes. We can disable a shard until it gets really disabled, not caring about service changes reset?

It looks like a kludge, because now the "configuration" includes both files on disk and some transient state.
We should also NOT write to the configuration, this defeats the purpose of the configuration.

The kludge is also not needed IMO, we do this purely to simplify usage in some scenarios.

It looks like a kludge, because now the "configuration" includes both files on disk and some transient state. We should also NOT write to the configuration, this defeats the purpose of the configuration. The kludge is also not needed IMO, we do this purely to simplify usage in some scenarios.
fyrchik added the
P1
label 2024-01-30 10:23:08 +00:00
fyrchik self-assigned this 2024-01-30 10:23:34 +00:00
fyrchik was unassigned by dstepanov-yadro 2024-01-30 13:26:55 +00:00
dstepanov-yadro self-assigned this 2024-01-30 13:26:56 +00:00
fyrchik self-assigned this 2024-01-30 13:28:00 +00:00
dstepanov-yadro was unassigned by fyrchik 2024-01-30 13:28:00 +00:00
fyrchik removed their assignment 2024-01-30 13:28:04 +00:00
dstepanov-yadro was assigned by fyrchik 2024-01-30 13:28:08 +00:00

See #945 description for details

See https://git.frostfs.info/TrueCloudLab/frostfs-node/pulls/945 description for details
Sign in to join this conversation.
No Milestone
No Assignees
4 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: TrueCloudLab/frostfs-node#917
There is no content yet.