IDS Forum

Re: Logical Log File not found ..

Posted By: JACQUES RENAUT
Date: Tuesday, 2 July 2013, at 2:06 p.m.

In Response To: Logical Log File not found .. (AKSHAY POLJI)

Original post:

Hello all,

An informix instance on Secondary node of a Hardware Cluster is refusing to come up and giving following error.

Wed Jun 26 04:10:30 2013

04:10:36 IBM Informix Dynamic Server Version 11.50.UC7 Software Serial Number AAA#B000000
04:10:37 Logical Log File not found.
04:10:37 oninit: Fatal error in shared memory initialization
04:10:37 IBM Informix Dynamic Server Stopped.
04:10:37 mt_shm_remove: WARNING: may not have removed all/correct segments

By Hardware Cluster I mean,

A virtual I/P to a Cluster which points to a Primary Node & upon Primary Node failure switches the package to Secondary Node. Also the Volume Group for raw devices is being shared and it's shifted to Secondary Node.

So the problem is whenever Cluster is pointing to Primary node, everything seems working fine. But when the package is shifted and Cluster points to Secondary node then the same Informix instance refuses to come up with above error.

Any body have any idea whats happening here ??

Regards,
Akshay Polji.

Response:

That error basically means it can't find the checkpoint record for where physical and logical recovery is supposed to start from, so the server can't come up. It seems like there could be some problem when you go to shift the volume group/raw devices from the primary to the secondary something is going wrong. The servers code does writes in a particular sequence to make sure that the checkpoint record would get to disk prior to reserve page changes, but we also assume when a write call returns, it's completed and made it to disk. If something could be preventing some of the database writes to make it to disk or do them out of order, you could see some sort of error like this. Also, are you sure all the disks are getting over the secondary? If for some reason your root chunk/dbspace wasn't shared properly or didn't get over to the secondary but there was some disk there that was accessible but didn't have the updated reserve pages you might see some problem like that.

You might want to do an oncheck -pr on your secondary and look at the checkpoint information for where the recovery checkpoint is and compare that to the logical log information. Then once you see the checkpoint location on the secondary, look at the last checkpoint completed message in MSGPATH of your primary, the checkpoint location should be either that checkpoint, or the checkpoint prior to that (in MSGPATH)

Example if you aren't sure what I mean:

info from oncheck -pr

Validating PAGE_1CKPT & PAGE_2CKPT...

Using check point page PAGE_1CKPT.

Time stamp of checkpoint 0x4b02de

Time of checkpoint 07/02/2013 12:48:49

Physical log begin address 1:263

Physical log size 25000 (p)

Physical log position at Ckpt 13595

Logical log unique identifier 56

Logical log position at Ckpt 0xafd018 (Page 2813, byte 24)

Checkpoint Interval 6335

So from that info on my system, the last checkpoint is logical log unique identifier 56, log position 0xafd018.... (more oncheck -pr output)

Log file number 2

Unique identifier 56

Log contains last checkpoint: Page 2813, byte 24

Log file flags 0x3 Log file in use

& Current log file

Physical location 1:30263

Log size 5000 (p)

So there I show that log file number 2 contains log unique identifier 56...so my log is present. On your system, I believe you will not find the log unique identifier (which is what that error means)

Now I'll show the message from MSGPATH for that completed checkpoint

12:48:49 Checkpoint Completed: duration was 0 seconds.
12:48:49 Tue Jul 2 - loguniq 56, logpos 0xafd018, timestamp: 0x4b02e1 Interval: 6335

So you can see the loguniq (that's log unique identifier) is 56 and the log pos (log position) 0xafd018.

Also, another thing you could check if possible is if you run oncheck -pr from the primary cluster before you switch over, that output should match the oncheck -pr output collected from your secondary cluster, and if it doesn't...you've got some sort of problem.

Jacques Renaut
IBM Informix Advanced Support
APD Team

Messages In This Thread

• Logical Log File not found ..
  AKSHAY POLJI -- Tuesday, 2 July 2013, at 1:32 p.m.
  • Re: Logical Log File not found ..
    JACQUES RENAUT -- Tuesday, 2 July 2013, at 2:06 p.m.
    • Re: Logical Log File not found ..
      AKSHAY POLJI -- Wednesday, 3 July 2013, at 6:16 a.m.
      • Re: Logical Log File not found ..
        JACQUES RENAUT -- Wednesday, 3 July 2013, at 9:12 a.m.
        • Re: Logical Log File not found ..
          AKSHAY POLJI -- Wednesday, 3 July 2013, at 11:21 a.m.
  • RE: Logical Log File not found ..
    Alexandre Marini -- Tuesday, 2 July 2013, at 2:34 p.m.
    • Re: RE: Logical Log File not found ..
      AKSHAY POLJI -- Wednesday, 3 July 2013, at 4:43 a.m.
  • Re: Logical Log File not found ..
    Art Kagel -- Tuesday, 2 July 2013, at 3:05 p.m.
    • Re: Logical Log File not found ..
      AKSHAY POLJI -- Wednesday, 3 July 2013, at 5:16 a.m.

IDS Forum is maintained by Administrator with WebBBS 5.12.