[sheepdog-users] 答复: sheepdog(v0.9.1) cluster will crash when a node running sheep v0.7.9 want to join the cluster.

[redtone]

I do not want to run v0.7.x and v0.9.x at the same time. 

But the mistake to run v.0.7.x will cause the production cluster running v0.
9.x crashed; this is big issue cause the cluster not available.

-----邮件原件-----
发件人: Hitoshi Mitake [mailto:mitake.hitoshi at gmail.com] 代表 Hitoshi Mitake
发送时间: 2015年1月20日 12:21
收件人: redtone
抄送: sheepdog-users at lists.wpkg.org
主题: Re: [sheepdog-users] sheepdog(v0.9.1) cluster will crash when a node
running sheep v0.7.9 want to join the cluster.

At Mon, 19 Jan 2015 21:26:23 +0800,
redtone wrote:
> 
> I run a sheepdog cluster with sheepdog v0.9.1 with zookeeper v3.4.6
> 
> When I run sheep command to add a node to the cluster, the cluster
crashed.
> 
> Then I found the node's sheep version is v 0.7.9.
> 
> How to prohibit such issue ?

Co-existing v0.7.x and v0.9.x isn't supported. Could you update entire
cluster to v0.9.1? In addition, store format is also incompatible. You
need to backup your VDIs with qemu-img and restore later. Sorry for
inconvenience.

Thanks,
Hitoshi

> 
> The following is the log message in sheep.log
> 
> Jan 19 20:58:24   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 20:58:56   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 20:59:28   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 20:59:28   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 20:59:28   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 21:00:00   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 21:00:00   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 21:00:32   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 21:00:32   INFO [main] zk_find_master(838) detect master leave,
start
> to compete master
> Jan 19 21:00:32  EMERG [main] crash_handler(268) sheep exits unexpectedly
> (Segmentation fault).
> Jan 19 21:00:32  EMERG [main] sd_backtrace(833) sheep.c:270: crash_handler
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847)
> /lib64/libpthread.so.0(+0xf70f) [0x7f9a1772a70f]
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847)
> /lib64/libc.so.6(memcpy+0x3e6) [0x7f9a16d6ad46]
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847)
> /lib/libzookeeper_mt.so.2(+0xc77c) [0x7f9a1728977c]
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847)
> /lib/libzookeeper_mt.so.2(+0xdbbd) [0x7f9a1728abbd]
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847)
> /lib/libzookeeper_mt.so.2(zoo_aset+0xdc) [0x7f9a17285cac]
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847)
> /lib/libzookeeper_mt.so.2(zoo_set2+0x68) [0x7f9a17285e18]
> Jan 19 21:00:32  EMERG [main] sd_backtrace(833) zookeeper.c:274:
zk_set_data
> Jan 19 21:00:32  EMERG [main] sd_backtrace(833) zookeeper.c:1252:
> zk_event_handler
> Jan 19 21:00:32  EMERG [main] sd_backtrace(833) event.c:210: do_event_loop
> Jan 19 21:00:32  EMERG [main] sd_backtrace(833) sheep.c:963: main
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847)
> /lib64/libc.so.6(__libc_start_main+0xfc) [0x7f9a16cffd1c]
> Jan 19 21:00:32  EMERG [main] sd_backtrace(847) sheep() [0x404238]
> 
> 
> 
> 
> 
> -- 
> sheepdog-users mailing lists
> sheepdog-users at lists.wpkg.org
> https://lists.wpkg.org/mailman/listinfo/sheepdog-users