Project

General

Profile

Actions

Bug #21828

closed

Crc check failure with Ceph Luminous on AArch64 platform

Added by Anonymous over 6 years ago. Updated almost 3 years ago.

Status:
Can't reproduce
Priority:
High
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
2 - major
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

Deployed Ceph Luminous (v12.2.0 and v12.2.1) on Arm server, the Ceph cluster status is "HEALTH_OK". There are 1 Mon and 3 OSDs.

Did some actions to validate the basic function:
  • add another OSD
  • deploy MDS
  • map rbd

All the above operations are failed although the cluster is still "HEALTH_OK".

From the log of ceph-mon, there are many crc error.
ubuntu@arm:~/tone/ceph-cluster$ sudo tail f /var/log/ceph/ceph-mon.arm02.log
2017-10-17 10:33:55.165485 ffff771e0a30 0 -
10.169.40.87:6789/0 >> - conn(0xaaaae34c1000 :6789 s=STATE_ACCEPTING_WAIT_BANNER_ADDR pgs=0 cs=0 l=0)._process_connection accept peer addr is really 10.169.40.87:0/3481051564 (socket is )
2017-10-17 10:33:55.165822 ffff771e0a30 0 -
10.169.40.87:6789/0 >> 10.169.40.87:0/3481051564 conn(0xaaaae34c1000 :6789 s=STATE_OPEN_MESSAGE_HEADER pgs=1 cs=1 l=1).process got bad header crc 0 != 3959462138
2017-10-17 10:34:05.598228 ffff771e0a30 0 -- 10.169.40.87:6789/0 >> 10.169.40.87:0/3481051564 conn(0xaaaae393c800 :6789 s=STATE_OPEN_MESSAGE_HEADER pgs=2 cs=1 l=1).process got bad header crc 0 != 3959462138
2017-10-17 10:34:15.614262 ffff771e0a30 0 -- 10.169.40.87:6789/0 >> 10.169.40.87:0/3481051564 conn(0xaaaae35c9000 :6789 s=STATE_OPEN_MESSAGE_HEADER pgs=3 cs=1 l=1).process got bad header crc 0 != 3959462138
2017-10-17 10:34:23.989030 ffff7d1e0a30 0 mon.arm02@0(leader).data_health(3) update_stats avail 87% total 795 GB, used 63049 MB, avail 693 GB
2017-10-17 10:34:25.598138 ffff771e0a30 0 -- 10.169.40.87:6789/0 >> 10.169.40.87:0/3481051564 conn(0xaaaae35c6000 :6789 s=STATE_OPEN_MESSAGE_HEADER pgs=4 cs=1 l=1).process got bad header crc 0 != 3959462138
2017-10-17 10:34:35.614253 ffff771e0a30 0 -- 10.169.40.87:6789/0 >> 10.169.40.87:0/3481051564 conn(0xaaaae35a5800 :6789 s=STATE_OPEN_MESSAGE_HEADER pgs=5 cs=1 l=1).process got bad header crc 0 != 3959462138
2017-10-17 10:34:45.598203 ffff771e0a30 0 -- 10.169.40.87:6789/0 >> 10.169.40.87:0/3481051564 conn(0xaaaae393c800 :6789 s=STATE_OPEN_MESSAGE_HEADER pgs=6 cs=1 l=1).process got bad header crc 0 != 3959462138

Then re-generated all the Debian installation packages with script "make-debs.sh" on Arm host, and re-deploy Ceph-cluster, everything is OK.

It is very strange that the crc value is zero. It seems the crc calculation is not activating on AArch64 platform.

The Linux dist is Ubuntu:
ubuntu@arm02:/var/lib$ uname -a
Linux arm02 4.4.0-96-generic #119-Ubuntu SMP Tue Sep 12 15:00:28 UTC 2017 aarch64 aarch64 aarch64 GNU/Linux

Actions #1

Updated by Anonymous over 6 years ago

  • Subject changed from Ceph doesn't work on AArch64 platform to Crc check failure with Ceph Luminous on AArch64 platform
Actions #2

Updated by Sage Weil almost 3 years ago

  • Status changed from New to Can't reproduce
Actions

Also available in: Atom PDF