TrueCloudLab/frostfs-node
20
1
Fork 27
Code Issues 100 Pull requests 4 Projects Releases 1 Activity Actions

node: Evacuate objects without setting mode to MAINTENANCE #1349

Merged
fyrchik merged 1 commit from acid-ant/frostfs-node:feature/evac-without-mm into master 2024-09-05 13:12:56 +00:00
Conversation 2 Commits 1 Files changed 18 +118 -10
acid-ant commented 2024-09-03 12:47:27 +00:00
Member
Copy link

Signed-off-by: Anton Nikiforov an.nikiforov@yadro.com

Signed-off-by: Anton Nikiforov <an.nikiforov@yadro.com>
acid-ant added 1 commit 2024-09-03 12:47:33 +00:00
[#xx] node: Evacuate objects without setting mode to MAINTENANCE
Some checks failed
DCO action / DCO (pull_request) Failing after 3m5s
Details
Tests and linters / Run gofumpt (pull_request) Successful in 3m3s
Details
Tests and linters / Tests (1.23) (pull_request) Successful in 4m45s
Details
Tests and linters / Tests (1.22) (pull_request) Successful in 4m47s
Details
Tests and linters / Lint (pull_request) Successful in 5m3s
Details
Pre-commit hooks / Pre-commit (pull_request) Successful in 5m12s
Details
Vulncheck / Vulncheck (pull_request) Successful in 4m57s
Details
Tests and linters / Staticcheck (pull_request) Successful in 5m39s
Details
Tests and linters / Tests with -race (pull_request) Successful in 6m47s
Details
Build / Build Components (1.23) (pull_request) Successful in 6m58s
Details
Build / Build Components (1.22) (pull_request) Successful in 8m34s
Details
Tests and linters / gopls check (pull_request) Successful in 8m50s
Details
d2c27341a9 
Signed-off-by: Anton Nikiforov <an.nikiforov@yadro.com>
acid-ant force-pushed feature/evac-without-mm from d2c27341a9 to 280c981cb2 2024-09-03 12:49:51 +00:00 Compare
acid-ant requested review from storage-core-committers 2024-09-03 13:44:31 +00:00
acid-ant requested review from storage-core-developers 2024-09-03 13:44:33 +00:00
fyrchik reviewed 2024-09-03 14:06:53 +00:00
pkg/local_object_storage/engine/evacuate.go Outdated
@ -366,6 +366,7 @@ func (e *StorageEngine) evacuateShardObjects(ctx context.Context, shardID string
listPrm.WithCount(defaultEvacuateBatchSize)
sh := shardsToEvacuate[shardID]
sh.SetEvacInProgress(true)
fyrchik commented 2024-09-03 14:00:24 +00:00
Owner
Copy link

When is this disabled? What if we get error in ListWithCursor below?

When is this disabled? What if we get error in `ListWithCursor` below?
acid-ant commented 2024-09-04 06:23:45 +00:00
Author
Member
Copy link

When we move shard to DISABLED mode. Think this should be the one-way trip. Otherwise, it is possible to lose the data.

When we move shard to `DISABLED` mode. Think this should be the one-way trip. Otherwise, it is possible to lose the data.
dstepanov-yadro commented 2024-09-04 07:30:41 +00:00
Member
Copy link

But for evacuation shard mode READ_ONLY is only required...

But for evacuation shard mode `READ_ONLY` is only required...
pkg/local_object_storage/shard/head.go Outdated
@ -70,6 +72,11 @@ func (s *Shard) Head(ctx context.Context, prm HeadPrm) (HeadRes, error) {
res, err = s.Get(ctx, getPrm)
obj = res.Object()
} else {
s.m.RLock()
fyrchik commented 2024-09-03 14:03:42 +00:00
Owner
Copy link

On an unrelated note: we should take mode for the whole operation duration.
I am surprised it wasn't the case before.
Not for this PR

On an unrelated note: we should take mode for the whole operation duration. I am surprised it wasn't the case before. Not for this PR
acid-ant commented 2024-09-04 06:06:03 +00:00
Author
Member
Copy link

That was done in a such way because we take the same lock here res, err = s.Get(ctx, getPrm). Maybe that was the reason why we don't take lock for the whole operation.

That was done in a such way because we take the same lock here `res, err = s.Get(ctx, getPrm)`. Maybe that was the reason why we don't take lock for the whole operation.
pkg/local_object_storage/shard/info.go Outdated
@ -17,2 +17,4 @@
Mode mode.Mode
// True when evacuation is in progress.
EvacInProgress bool
fyrchik commented 2024-09-03 14:01:36 +00:00
Owner
Copy link

EvacuationInProgress?

`EvacuationInProgress`?
acid-ant commented 2024-09-04 06:30:19 +00:00
Author
Member
Copy link

Renamed.

Renamed.
pkg/local_object_storage/shard/shard.go Outdated
@ -580,2 +580,4 @@
}
}
func (s *Shard) SetEvacInProgress(val bool) {
fyrchik commented 2024-09-03 14:06:52 +00:00
Owner
Copy link

Evacuation is done on the engine level, but we have a flag on the shard level and because of this we have SkipEvacCheck parameter to the operation.
Was having evacuationInProgress flag considered to be present somewhere in the engine structure?

Evacuation is done on the engine level, but we have a flag on the shard level _and_ because of this we have `SkipEvacCheck` parameter to the operation. Was having `evacuationInProgress` flag considered to be present somewhere in the engine structure?
acid-ant commented 2024-09-04 07:03:18 +00:00
Author
Member
Copy link

It looks more complex and slower. Once it will be on an engine level, we should check for each request is evacuation for shard in progress, and go to internal map for checking shard and exclude it from iteration. If this overhead is ok for us, we can do that in a such way.

It looks more complex and slower. Once it will be on an `engine` level, we should check for each request is evacuation for shard in progress, and go to internal map for checking shard and exclude it from iteration. If this overhead is ok for us, we can do that in a such way.
acid-ant force-pushed feature/evac-without-mm from 280c981cb2 to 785403affd 2024-09-04 06:30:10 +00:00 Compare
acid-ant force-pushed feature/evac-without-mm from 785403affd to 659c5989ce 2024-09-04 13:07:13 +00:00 Compare
acid-ant commented 2024-09-04 13:09:47 +00:00
Author
Member
Copy link

Propagate evacuation status into metrics:

# HELP frostfs_node_engine_evacuation_in_progress Shard evacuation in progress
# TYPE frostfs_node_engine_evacuation_in_progress gauge
frostfs_node_engine_evacuation_in_progress{mode="false",shard_id="ALW7GShy8aBd8CyG9VUKdW"} 1

Now evacuation status resetting by request StopShardEvacuationRequest.

Propagate evacuation status into metrics: ``` # HELP frostfs_node_engine_evacuation_in_progress Shard evacuation in progress # TYPE frostfs_node_engine_evacuation_in_progress gauge frostfs_node_engine_evacuation_in_progress{mode="false",shard_id="ALW7GShy8aBd8CyG9VUKdW"} 1 ``` Now evacuation status resetting by request `StopShardEvacuationRequest`.
acid-ant force-pushed feature/evac-without-mm from 659c5989ce to 5717d86173 2024-09-04 13:11:08 +00:00 Compare
dstepanov-yadro approved these changes 2024-09-04 13:28:41 +00:00
Dismissed
fyrchik reviewed 2024-09-05 08:08:39 +00:00
fyrchik left a comment
Owner
Copy link

Please, adjust the documentation https://git.frostfs.info/TrueCloudLab/frostfs-node/src/branch/master/docs/evacuation.md (need to mention new flag and the behaviour)

Please, adjust the documentation https://git.frostfs.info/TrueCloudLab/frostfs-node/src/branch/master/docs/evacuation.md (need to mention new flag and the behaviour)
acid-ant force-pushed feature/evac-without-mm from 5717d86173 to fb01817d3d 2024-09-05 13:07:12 +00:00 Compare
acid-ant dismissed dstepanov-yadro's review 2024-09-05 13:07:12 +00:00
Reason:

New commits pushed, approval review dismissed automatically according to repository settings

acid-ant force-pushed feature/evac-without-mm from fb01817d3d to 108e4e07be 2024-09-05 13:08:46 +00:00 Compare
acid-ant commented 2024-09-05 13:11:01 +00:00
Author
Member
Copy link

Updated evacuation.md.

Updated `evacuation.md`.
fyrchik approved these changes 2024-09-05 13:12:51 +00:00
fyrchik merged commit 108e4e07be into master 2024-09-05 13:12:56 +00:00
Sign in to join this conversation.
Reviewers
No reviewers
fyrchik
TrueCloudLab/storage-core-developers
Labels
Clear labels   
P0

Highest

  
P1

High

  
P2

Medium

  
P3

Low

  
badger

Badger DB related issues

  
frostfs-adm

   
frostfs-cli

   
frostfs-ir

   
frostfs-lens

   
frostfs-node

   
good first issue

Good for newcomers

  
triage
  
Infrastructure

Infrastructure, Automation, CI/CD or DevOps related stuff

  
blocked

The issue or PR is blocked by something

  
bug

Something is not working

  
config

Configuration format update or breaking change

  
discussion

Talking about something in order to reach a decision or to exchange ideas

  
documentation

Documentation related

  
duplicate

This issue or pull request already exists

  
enhancement

New feature

  
go

Language features adoption

  
help wanted

Extra attention is needed

  
internal

   
invalid

Something is wrong

  
kludge

This is a kludge and we know it

  
observability

Related to metrics, tracing and logs

  
perfomance

Perfomance related

  
question

More information is needed

  
refactoring

Refactoring

  
wontfix

This won't be fixed

No labels
P0
P1
P2
P3
badger
frostfs-adm
frostfs-cli
frostfs-ir
frostfs-lens
frostfs-node
good first issue
triage
Infrastructure
blocked
bug
config
discussion
documentation
duplicate
enhancement
go
help wanted
internal
invalid
kludge
observability
perfomance
question
refactoring
wontfix
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
3 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#1349
No description provided.
Delete branch "acid-ant/frostfs-node:feature/evac-without-mm"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?