This discussion is locked
etblack

Q: xsan panic and recurring errors

I had the volume go down in a panic last week. We were able to restart and get everything running but now have a persistent error which appears to be related. Both reference gethostbyname. I have dns setup for all of the systems on both the regular network and the metadata. One of the issues is that there is no secondary mdc, a problem the client refuses to rectify.

 

 

 

 

Process:         fsm [189]

Path:            /Library/Filesystems/Xsan/bin/fsm

Identifier:      fsm

Version:         ??? (???)

Code Type:       X86-64 (Native)

Parent Process:  fsmpm [179]

 

 

Date/Time:       2012-03-26 13:58:00.591 -0700

OS Version:      Mac OS X Server 10.6.8 (10K549)

Report Version:  6

 

 

Exception Type:  EXC_CRASH (SIGABRT)

Exception Codes: 0x0000000000000000, 0x0000000000000000

Crashed Thread:  0  Dispatch queue: com.apple.main-thread

 

 

Application Specific Information:

PANIC: /Library/Filesystems/Xsan/bin/fsm "Server_comm_init GetHostByName failed" file server_comm.c, line 3455

 

 

 

Thread 0 Crashed:  Dispatch queue: com.apple.main-thread

0   libSystem.B.dylib                       0x00007fff848529da __pthread_kill + 10

1   libSystem.B.dylib                       0x00007fff848522fe pthread_kill + 83

2   fsm                                     0x00000001000ac22c 0x100000000 + 705068

3   fsm                                     0x00000001000ac787 0x100000000 + 706439

4   fsm                                     0x00000001000ea56a 0x100000000 + 959850

5   fsm                                     0x00000001000767f9 0x100000000 + 485369

6   fsm                                     0x0000000100030aa4 0x100000000 + 199332

7   fsm                                     0x0000000100000d94 0x100000000 + 3476

 

 

This error keeps cropping up in the current logs:

 

 

prod servermgrd[16235]: xsan: [16235/2112B0] ERROR: get_fsmvol_at_index: Could not connect to FSM because Cannot get host by name - No such file or directory

Posted on Apr 3, 2012 11:02 AM

Close

Q: xsan panic and recurring errors

  • All replies
  • Helpful answers

  • by Julius.J,

    Julius.J Julius.J Apr 3, 2012 5:04 PM in response to etblack
    Level 1 (0 points)
    Apr 3, 2012 5:04 PM in response to etblack

    This is the log form the MDC?
     
    Can you look at the fsnameservers file in the xsan preferences directory and see if the mdc's ip is the only one on the list.
    library/filesystems/xsan/fsnameservers (I think I'm on 10.7 now and they moved it).
     
    Can you run sudo changeip -checkhostname on the MDC and post the output (it should just be the ip address of the MDC and its dns name/hostname).

     

    *Edit oh this is the same post~

  • by Blaidd Drwg,

    Blaidd Drwg Blaidd Drwg Apr 4, 2012 8:14 PM in response to etblack
    Level 1 (109 points)
    Apr 4, 2012 8:14 PM in response to etblack

    Verify your MDC gets an answer when doing a reverse lookup on each of its IPs. All DNS servers should respond with the same answer for each IP address. If they don't provide an answer or there's a mismatch, fix that in the DNS server.

     

    Alternatively, you could simply add lines in /etc/hosts on the MDC, like this for the metadata IP address:

     

    192.168.1.1 mdc1-meta.example.com

  • by etblack,

    etblack etblack Apr 5, 2012 9:53 AM in response to Julius.J
    Level 1 (0 points)
    Photos for Mac
    Apr 5, 2012 9:53 AM in response to Julius.J

    I've checked with changeip -checkhostname and the names match. I've done an nslookup on the mdc and client's ips both on the standard network on the metadata network and they match. The fsnameservers shows the metadata ip of the MDC (which is running the dns), but no hostnames.

  • by Blaidd Drwg,

    Blaidd Drwg Blaidd Drwg Apr 6, 2012 12:29 AM in response to etblack
    Level 1 (109 points)
    Apr 6, 2012 12:29 AM in response to etblack

    Hmm... that crash should only occur when there's a DNS failure of some kind. Try contacting AppleCare.

  • by Julius.J,

    Julius.J Julius.J Apr 6, 2012 3:31 PM in response to etblack
    Level 1 (0 points)
    Apr 6, 2012 3:31 PM in response to etblack

    It might be worth while to clear your dns cache on the mdc if its also the dns.

     

    ~dscacheutil -flushcache

     

    Is there anything else in the logs that seem relevent?

    Have you run a permissions fix on the mdc anytime recently?

     

    In your dns setting you have one host name for the public and one for the private?

    internet - 192.168.1.100 mdc01."yourdomain"."internal"

    private net - 192.168.2.100 mdc01-san."yourdomain"."internal"

     

    Do you have any crostalk? (if only the private network is pluged in does it get a dhcp address? If you add the dns address to your static private port it can you get online?)

  • by Blaidd Drwg,Solvedanswer

    Blaidd Drwg Blaidd Drwg Apr 9, 2012 10:39 AM in response to Julius.J
    Level 1 (109 points)
    Apr 9, 2012 10:39 AM in response to Julius.J

    I know you said DNS looks good. But try adding the /etc/hosts entries for the MDC's public and private IPs anyway as a test. Do the GetHostByName panics continue? I bet they stop.

  • by etblack,

    etblack etblack Apr 10, 2012 2:13 PM in response to Blaidd Drwg
    Level 1 (0 points)
    Photos for Mac
    Apr 10, 2012 2:13 PM in response to Blaidd Drwg

    I added the hosts to /etc/host and then flushed the cache and it seems to be working now. No more errors in the log.