Project

General

Profile

Actions
Copy link

Bug #21008

closed

clone flatten is pending in 4% when it uses ec pool

Added by Tang Jin over 6 years ago. Updated over 6 years ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Target version:
Ceph - v12.2.0
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
1 - critical
Reviewed:
Affected Versions:
Ceph - v12.1.0
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

If flatten a clone when pool is ec pool, it will be pending in 4%, and I check the client log, it read a data object from ec data pool and can't return. Meanwhile I check the replicated pool size is 2, and I am sure that data object exist in ec data pool.
But if it set replicated pool size to 1, flatten will succeed.

Files

long_text_2017-08-22.txt (81.9 KB) long_text_2017-08-22.txt Tang Jin, 08/22/2017 02:13 AM
Actions
Copy link
 #1

Updated by Tang Jin over 6 years ago

If flatten a clone when pool is ec pool, it will hang in 4%, and I check the client log, it read a data object from ec data pool and can't return. Meanwhile I check the replicated pool size is 2, and I am sure that data object exist in ec data pool.
If the cluster has only one host, replicated pool size is 1, it can finish flatten.

Actions
Copy link
 #2

Updated by Greg Farnum over 6 years ago

  • Project changed from Ceph to rbd
  • Category deleted (OSD)
Actions
Copy link
 #3

Updated by Tang Jin over 6 years ago

@Greg Farnum Farnum
can ceph rbd support this function (clone flatten from a ec pool)?

Actions
Copy link
 #4

Updated by Venky Shankar over 6 years ago

Tang Jin wrote:

@Greg Farnum Farnum
can ceph rbd support this function (clone flatten from a ec pool)?

I'll try to recreate. Meanwhile, could you provide log (with "debug rbd = 20") for the problematic run?

Actions
Copy link
 #5

Updated by Tang Jin over 6 years ago

@Greg Farnum Farnum
here is rbd flatten cmd hung log named "long_text_2017-08-22.txt"

Actions
Copy link
 #6

Updated by Venky Shankar over 6 years ago

Tang Jin wrote:

@Greg Farnum Farnum
here is rbd flatten cmd hung log named "long_text_2017-08-22.txt"

what's the cluster health (ceph -s)?

so the flatten operation never progresses at all? how long was this observed for?

Actions
Copy link
 #7

Updated by Jason Dillaman over 6 years ago

  • Status changed from New to Need More Info

@Tang: the attached log shows that librbd is waiting for a response from the OSDs. Can you re-run with "--debug-rbd=20 --debug-objecter=20" added as CLI arguments?

Actions
Copy link
 #8

Updated by Jason Dillaman over 6 years ago

  • Status changed from Need More Info to Closed

Closing this for now. If the requested information can be provided, please feel free to re-open.

Actions
Copy link

Also available in: Atom PDF