TrueCloudLab/frostfs-node
18
1
Fork 27
Code Issues 94 Pull requests 3 Projects Releases 1 Activity Actions

ir: Add health status reporting on reconfiguration #1311

Merged
fyrchik merged 1 commit from elebedeva/frostfs-node:fix/ir-reload-notify-systemd into master 2024-09-04 19:51:11 +00:00
Conversation 0 Commits 1 Files changed 3 +13 -3
elebedeva commented 2024-08-14 11:28:59 +00:00
Member
Copy link

Fix #1135
Relates to #1262

frostfs-ir service reconfigures correctly and service's statuses are being reported to systemd. However, since we replaced go:linkname & nanotime() with time.Since(), systemd refuses to accept reload signal response from frostfs-ir. To maintain correct behaviour it was decided to revevrt systemd-related changes until a better solution is found.

Signed-off-by: Ekaterina Lebedeva ekaterina.lebedeva@yadro.com

Fix #1135 Relates to #1262 `frostfs-ir` service reconfigures correctly and service's statuses are being reported to systemd. However, since we replaced `go:linkname` & `nanotime()` with `time.Since()`, systemd refuses to accept reload signal response from `frostfs-ir`. To maintain correct behaviour it was decided to revevrt[ systemd-related changes](https://git.frostfs.info/elebedeva/frostfs-node/commit/327d364f34de730879f330ea51d8801f5c6bddc9) until a better solution is found. Signed-off-by: Ekaterina Lebedeva <ekaterina.lebedeva@yadro.com>
elebedeva added the
frostfs-ir
internal
 labels 2024-08-14 11:28:59 +00:00
elebedeva requested review from storage-core-committers 2024-08-14 11:29:54 +00:00
elebedeva requested review from storage-core-developers 2024-08-14 11:29:58 +00:00
fyrchik reviewed 2024-08-14 11:43:18 +00:00
go.mod Outdated
@ -45,2 +44,2 @@
google.golang.org/grpc v1.63.2
google.golang.org/protobuf v1.33.0
golang.org/x/term v0.21.0
google.golang.org/grpc v1.65.0
fyrchik commented 2024-08-14 11:39:41 +00:00
Owner
Copy link

no, too early #1268

no, too early https://git.frostfs.info/TrueCloudLab/frostfs-node/issues/1268
elebedeva commented 2024-08-15 13:56:05 +00:00
Author
Member
Copy link

fixed

fixed
fyrchik marked this conversation as resolved
pkg/util/sdnotify/sdnotify.go Outdated
@ -17,7 +16,6 @@ const (
var (
socket *net.UnixAddr
start = time.Now()
fyrchik commented 2024-08-14 11:40:09 +00:00
Owner
Copy link

Have you tried using start time.Time (i.e. the default value)?

Have you tried using `start time.Time` (i.e. the default value)?
elebedeva commented 2024-08-15 11:24:15 +00:00
Author
Member
Copy link

The time.Time default value is 0001-01-01 00:00:00 +0000 UTC, time.Since() returns time.Duration which is an alias of int64. According to Go doc:

The representation limits the largest representable duration to approximately 290 years.

time.Since(time.Time{}) overflows.

The `time.Time` default value is `0001-01-01 00:00:00 +0000 UTC`, `time.Since()` returns `time.Duration` which is an alias of `int64`. According to [Go doc](https://pkg.go.dev/time#Duration): > The representation limits the largest representable duration to approximately 290 years. `time.Since(time.Time{})` overflows.
fyrchik marked this conversation as resolved
pkg/util/sdnotify/sdnotify.go
@ -53,7 +51,7 @@ func FlagAndStatus(status string) error {
// must be sent, containing "READY=1".
//
// For MONOTONIC_USEC format refer to https://www.man7.org/linux/man-pages/man3/sd_notify.3.html
fyrchik commented 2024-08-14 11:43:15 +00:00
Owner
Copy link

Maybe in IR SIGHUP is so fast that we send the same MONOTONIC_USEC? This would've explained the problem.

Maybe in IR SIGHUP is so fast that we send the same `MONOTONIC_USEC`? This would've explained the problem.
elebedeva commented 2024-08-15 11:50:14 +00:00
Author
Member
Copy link

Receiving the same MONOTONIC_USEC doesn't seem like a problem to systemd. I tried sending time.Since(time.Time{}) (always the same value), systemd is OK with it, service reload is successful.

It appears as if systemd does not accept values less than some minimum and greater than some maximum, and time in us since start is deemed as being too small. Time in ns since start works fine most of the time but not always (i've got hang-ups a couple times). Haven't found those min & max values in systemd source code yet.

Passing a math.MaxInt64 as MONOTONIC_USEC works fine but math.MaxUint64 is not accepted.

Receiving the same `MONOTONIC_USEC` doesn't seem like a problem to `systemd`. I tried sending `time.Since(time.Time{})` (always the same value), `systemd` is OK with it, service reload is successful. It appears as if `systemd` does not accept values less than some minimum and greater than some maximum, and time in us since start is deemed as being too small. Time in ns since start works fine most of the time but not always (i've got hang-ups a couple times). Haven't found those min & max values in systemd source code yet. Passing a `math.MaxInt64` as `MONOTONIC_USEC` works fine but `math.MaxUint64` is not accepted.
fyrchik marked this conversation as resolved
elebedeva force-pushed fix/ir-reload-notify-systemd from bb3945f62a to 5da41f1fe5 2024-08-15 13:55:17 +00:00 Compare
fyrchik approved these changes 2024-08-15 13:57:23 +00:00
fyrchik merged commit 5da41f1fe5 into master 2024-08-15 14:00:10 +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
2 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#1311
No description provided.
Delete branch "elebedeva/frostfs-node:fix/ir-reload-notify-systemd"

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?