parallelvirtualcluster/pvc
1
0
Fork 0
Code Issues 11 Pull Requests Projects Releases 115 Wiki Activity

[FEATURE] Flush on daemon shutdown #3

New Issue
Closed
opened 2018-06-21 22:34:47 -04:00 by JoshuaBoniface · 14 comments
JoshuaBoniface commented 2018-06-21 22:34:47 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

Flush node on daemon shutdown, however also have a separate "restart" signal from Systemd trapped to do a graceful restart without flushing.

Suggested by @michal

Flush node on daemon shutdown, however also have a separate "restart" signal from Systemd trapped to do a graceful restart without flushing. Suggested by @michal
JoshuaBoniface commented 2018-06-21 22:35:05 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

changed the description

changed the description
JoshuaBoniface commented 2018-06-25 23:05:51 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

mentioned in commit d0dc923d92

mentioned in commit d0dc923d9201151b68e23c8b5210cb98a74b5ab1
JoshuaBoniface commented 2018-06-25 23:13:17 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

mentioned in commit ac2b3c464c

mentioned in commit ac2b3c464cff4643b1078b02c7024bc1d15e2a6c
JoshuaBoniface commented 2018-06-25 23:17:16 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

mentioned in commit c5a7ac0738

mentioned in commit c5a7ac073849d580c9adb71aa00d9302fb30d5aa
JoshuaBoniface commented 2018-06-26 12:09:16 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

I did some testing and based on how Systemd works, there is no way to selectively flush only on a stop - the restart does a stop + start unconditionally so there's no way to differentiate the two. Trying to do this with SIGHUP/reload also does not work since that doesn't respawn.

Won't add; admin is always responsible for flushing first.

cc: @MichalKozanecki

I did some testing and based on how Systemd works, there is no way to selectively flush only on a `stop` - the `restart` does a `stop` + `start` unconditionally so there's no way to differentiate the two. Trying to do this with `SIGHUP`/`reload` also does not work since that doesn't respawn. Won't add; admin is always responsible for flushing first. cc: @MichalKozanecki
JoshuaBoniface commented 2018-06-26 12:09:17 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

closed

closed
JoshuaBoniface commented 2018-06-26 12:10:40 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

mentioned in commit 9eb1855add

mentioned in commit 9eb1855add1cf1e988f48233cb161cc0ecd1708a
JoshuaBoniface commented 2018-06-26 12:10:40 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

mentioned in commit 75ed61908a

mentioned in commit 75ed61908a3315cabee1411818f97afd4f145298
JoshuaBoniface commented 2018-06-26 12:10:40 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

mentioned in commit f27cdb024c

mentioned in commit f27cdb024c9f83f2cf68029edcd91ac386b14380
MichalKozanecki commented 2018-06-26 21:38:45 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

How do the other nodes know the difference between a node being down because an admin gracefully shut pvc down, and pvc(or zookeeper) crashed?

How do the other nodes know the difference between a node being down because an admin gracefully shut `pvc` down, and `pvc`(or `zookeeper`) crashed?
MichalKozanecki commented 2018-06-26 21:38:45 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

reopened

reopened
JoshuaBoniface commented 2018-06-26 21:40:08 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

If pvc crashes, it never sets its daemon state to stop, and hence gets fenced. If Zookeeper crashes, the node stops sending keepalives, and hence gets fenced.

If `pvc` crashes, it never sets its daemon state to `stop`, and hence gets fenced. If Zookeeper crashes, the node stops sending keepalives, and hence gets fenced.
MichalKozanecki commented 2018-06-26 21:54:13 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

💯 👍

:100: :+1:
MichalKozanecki commented 2018-06-26 21:54:14 -04:00 (Migrated from git.bonifacelabs.ca)
Copy Link

closed

closed
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
API
blocked
Issue blocked by another issue
bug
CLI
Client
Issues abut the client frontends
Daemon
Issues abut the node daemon
debt
Technical debt that should be investigated or cleaned up
documentation
feature
improvement
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
joshuaboniface
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: parallelvirtualcluster/pvc#3
No description provided.
Delete Branch "%!s()"

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?