| Author |
Message
|
| sunny_30 |
 Posted: Tue Jul 14, 2009 9:00 am Post subject: |
 |
|
Master
Joined: 03 Oct 2005
Posts: 258
|
| Code: |
| This is nothing to do with WMQ or queue manager logs, but simple file handling and why people used shared disc to ensure consistency. |
Thank you Vitor! this answers my question.
| Code: |
| if you want to know what happens to in-flight transactions when you copy the in-use logs, why not just try it? Easy enough to set up a test rig. |
I tried & have seen it in DR(testing). The QM has not come up, so I had to manually recreate everything.
But we had another scenario in DR where "synchronous package replication" is done from PROD to DR box. For this, I have seen QM has come up fine on DR. May be it has to do something with replication being "synchronous" (or) there werent any transactions underway when active-logs got copied over.
| Code: |
| You also seem to be obsessed with this question of in-flight transactions to the exclusion of other issues |
correct- I tried this: While MQ-put of a 100-mb message is underway, I abruptly shut off the QM/laptop. The QM started fine.
My obsession was to know WHY the DR-QM doesnt come up (on some occassions) & the MQ /QM on same system comes up fine.
I think I now know the answer...
I still have trouble how to put my Question in words.. sorry abt the lengthy explanations.
| Code: |
| I suggest you go and look in the qm.ini file of an HA queue manager |
This implies the guy whoever did HA implementation here has not used MC91. Exerk, Thank you for enlightening.
We have a simple HACMP setup (at AIX level) with all MQ-file-systems shared except for mq-install (/usr/mqm). I will read the MC91 doc to see how HA is setup at MQ-level.
start-up scripts in /usr/sbin/cluster/local (AIX) control on which node the MQ comes up. Our failover setup is basic valid for any other application-not just MQ. I see other extra steps need to be undertaken for setting up MQ HA. |
|
| Back to top |
|
 |
| exerk |
| Posted: Tue Jul 14, 2009 12:58 pm Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| sunny_30 wrote: |
...This implies the guy whoever did HA implementation here has not used MC91. Exerk, Thank you for enlightening.
We have a simple HACMP setup (at AIX level) with all MQ-file-systems shared except for mq-install (/usr/mqm). I will read the MC91 doc to see how HA is setup at MQ-level.
start-up scripts in /usr/sbin/cluster/local (AIX) control on which node the MQ comes up. Our failover setup is basic valid for any other application-not just MQ. I see other extra steps need to be undertaken for setting up MQ HA. |
I would be really, really interested in the symlinks you find at /var/mqm/qmgrs level, and /var/mqm/logs, so please, if you would, do let me know - thank you.
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| sunny_30 |
| Posted: Tue Jul 14, 2009 7:26 pm Post subject: |
|
|
Master
Joined: 03 Oct 2005
Posts: 258
|
Hi Exerk,
There are no symlinks at both /var/mqm/log & /var/mqm/qmgrs locations.
In production, we have many times successfully failed over MQ & WMB applications.
-Sunny |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Tue Jul 14, 2009 7:49 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20763
Location: LI,NY
|
| sunny_30 wrote: |
Hi Exerk,
There are no symlinks at both /var/mqm/log & /var/mqm/qmgrs locations.
In production, we have many times successfully failed over MQ & WMB applications.
-Sunny |
So you don't really have HA. What you have is a shared mounting point(/var/mqm)? (contact admin) across 2 systems. Who is governing which system starts which qmgr?
Who is preventing both systems to attempt to start the same qmgr? 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| sunny_30 |
| Posted: Tue Jul 14, 2009 8:10 pm Post subject: |
|
|
Master
Joined: 03 Oct 2005
Posts: 258
|
Hi Saper,
We do have HA setup at AIX level.
The symlinks & startup/shutdown scripts are managed from /usr/sbin/cluster/local which control on what node the MQ & WMB file-systems will be mounted to. We have separate installs for MQ & WMB on both active & passive nodes. Only the working directories switch to the active node. After failover, the alias will point to the active system.
Im not exactly sure abt the working of MC91, but I believe that is implemented differently & may give more options to control MQ-HA
Thanks. |
|
| Back to top |
|
|
| exerk |
| Posted: Tue Jul 14, 2009 11:09 pm Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| sunny_30 wrote: |
| ...We do have HA setup at AIX level... |
No, you don't. You have, as fjb_saper points out, a shared mount point. When you apply maintenance to WMQ on the node that the /var/mqm file system is not on, what happens? I'd be interested in you trying.
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Jul 15, 2009 12:39 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| sunny_30 wrote: |
Im not exactly sure abt the working of MC91, but I believe that is implemented differently & may give more options to control MQ-HA
|
No, it implements HA properly. As my most worthy associates point out, you do not have HA you have a shared mount point. This is not the same thing and is a problem waiting to happen, for the reasons stated and others.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| sunny_30 |
| Posted: Wed Jul 15, 2009 4:23 am Post subject: |
|
|
Master
Joined: 03 Oct 2005
Posts: 258
|
On the system where MQ /WMB is currently running, these are the file-systems:--
active-node:
| Code: |
(mqm)$ df -g | egrep 'mqm|mqsi'
/dev/usrmqmlv 0.50 0.35 30% 2306 3% /usr/mqm
/dev/optmqsilv 1.50 0.74 51% 2452 2% /opt/IBM/mqsi
/dev/varmqmlv 10.00 9.79 3% 714 1% /var/mqm
/dev/QMNAECTP1_log 10.00 9.81 2% 9 1% /var/mqm/log/QMNAECTP1
/dev/qmnahubp2_log 12.50 11.25 11% 26 1% /var/mqm/log/QMNAHUBP2
/dev/QMNAECTP1_qmg 3.00 2.73 10% 305 1% /var/mqm/qmgrs/QMNAECTP1
/dev/qmnahubp2_qmgrs 12.50 12.04 4% 462 1% /var/mqm/qmgrs/QMNAHUBP2
/dev/varmqsilv 1.00 0.99 2% 446 1% /var/mqsi |
failover-node:
| Code: |
(mqm)$ df -g | egrep 'mqm|mqsi'
/dev/usrmqmlv 0.50 0.35 30% 2313 3% /usr/mqm
/dev/optmqsilv 1.50 0.76 50% 2455 2% /opt/IBM/mqsi |
| Code: |
| When you apply maintenance to WMQ on the node that the /var/mqm file system is not on, what happens? I'd be interested in you trying |
What does MQ-install have to do with /var/mqm ?
All files will be applied to /usr/mqm
| Code: |
| This is not the same thing and is a problem waiting to happen |
Any example of the problem please? |
|
| Back to top |
|
|
| exerk |
| Posted: Wed Jul 15, 2009 4:34 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| sunny_30 wrote: |
| exerk wrote: |
| When you apply maintenance to WMQ on the node that the /var/mqm file system is not on, what happens? I'd be interested in you trying |
What does MQ-install have to do with /var/mqm ?
All files will be applied to /usr/mqm |
Somebody please pinch me and tell me I am NOT seeing what is above...
...do you have any understanding of the installation paths of WebSphere MQ? Have you read the Quick Beginnings Guide?
I'm just going to sit back and watch the fireworks as you've now no doubt lit the blue touch-paper...
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| sunny_30 |
| Posted: Wed Jul 15, 2009 6:14 am Post subject: |
|
|
Master
Joined: 03 Oct 2005
Posts: 258
|
Exerk, it seems to me that you cannot think of MQ-HA other than those of MC91 terms. It is always good to use something that is exclusively designed for MQ but it doesnt make sense to assume HA cannot be setup in any other 'simpler' way. Im certainly open to learn where my setup falls short & what extra benefit we get from using MC91, before that I know I have to read the supportpac.
I have several times installed MQ with success when ONLY /usr/mqm file-system was available on AIX and also when ONLY /opt/mqm file-system was available on Linux.
The local directory structures under /var/mqm shd also be present but shared file-systems dont need to be mounted
May be you want to try the same?
Im willing to learn new things but you dont seem to be interested at all in anything that conflicts your way of reasoning. Its good that you want to have some fun the forum, but I wd go to an amusement park for fireworks.
If you dont feel it as too much burden (as if like teaching me the whole MQ-fundamentals-redbook, explaining me each bit line by line until I learn) for you to share your MQ-knowledge, you are welcome to do it. I wd certainly appreciate it but no problem even if you are not willing to get out of your "play-mode" |
|
| Back to top |
|
|
| exerk |
| Posted: Wed Jul 15, 2009 6:32 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| sunny_30 wrote: |
| ...it seems to me that you cannot think of MQ-HA other than those of MC91 terms... |
Because as far as I am concerned, that is the only way to think of it. Note that the MC91 SupportPac is provided by WebSphere MQ Development, Hursley, and if that group of people considered your method of HA to be viable, I'm sure it would be promoted by them, thereby saving them an awful lot of unnecessary work.
| sunny_30 wrote: |
| ...it doesnt make sense to assume HA cannot be setup in any other 'simpler' way... |
There's simple, and there's dangerous, and I consider what you are doing is dangerous.
| sunny_30 wrote: |
| ...I have several times installed MQ with success when ONLY /usr/mqm was available on AIX and also when ONLY /opt/mqm was available on Linux... |
Note: I didn't say install, I said apply maintenance.
| sunny_30 wrote: |
| ...I'm willing to learn new things but you dont seem to be interested at all in anything that conflicts your way of reasoning... |
I'm always willing to learn, that's why I like WMQ so much, because just when I think I've got a handle on it, I find out that I actually know so little of it - but one thing I did learn (early) was not to do HA as you have done it.
| sunny_30 wrote: |
| ...If you dont think its too much burden (like teaching me the whole MQ-fundamentals-redbook, explaining me each bit line by line until I learn) for you to share your MQ-knowledge, you are welcome to do it. I wd certainly appreciate it but no problem even if you are not willing to... |
I'm immune to sarcasm, but it's good to see that you're not scared to bite back - keep it up 
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| sunny_30 |
| Posted: Wed Jul 15, 2009 6:46 am Post subject: |
|
|
Master
Joined: 03 Oct 2005
Posts: 258
|
| Code: |
| Note: I didn't say install, I said apply maintenance. |
yes I did apply maintenance recently on a Linux failover system:
| Code: |
-bash-3.2$ uname -a
Linux atlsdbal008b 2.6.18-92.el5 #1 SMP Tue Apr 29 13:16:15 EDT 2008 x86_64 x86_64 x86_64 GNU/Linux
-bash-3.2$ dspmqver
Name: WebSphere MQ
Version: 6.0.2.6
CMVC level: p600-206-090217
BuildType: IKAP - (Production)
-bash-3.2$ df -a | grep mqm
-bash-3.2$ ls -ltr /var/mqm
total 44
drwxrwsrwx 2 mqm mqm 4096 Jun 3 22:12 trace
-rw-rw-r-- 1 mqm mqm 1373 Jun 3 22:12 service.env
drwxrwsr-x 3 mqm mqm 4096 Jun 3 22:12 qmgrs
-rw-rw-r-- 1 mqm mqm 2104 Jun 3 22:12 mqs.ini
drwxrwsr-x 2 mqm mqm 4096 Jun 3 22:12 log
drwxrwsr-x 2 mqm mqm 4096 Jun 3 22:12 exits64
drwxrwsr-x 2 mqm mqm 4096 Jun 3 22:12 exits
drwxrwsrwx 2 mqm mqm 4096 Jun 3 22:12 errors
drwxrwsr-x 3 mqm mqm 4096 Jun 3 22:12 conv
drwxrwsr-x 2 mqm mqm 4096 Jun 3 22:12 config
drwxr-sr-x 2 root mqm 4096 Jun 3 22:15 tivoli
-bash-3.2$ ls -ltr /var/mqm/log
total 0
-bash-3.2$ ls -ltr /var/mqm/qmgrs
total 4
drwxrwsr-x 2 mqm mqm 4096 Jun 3 22:12 @SYSTEM |
This is what we have on primary:
| Code: |
(atlsdbal008a)/>uname -a
Linux atlsdbal008a 2.6.18-92.el5 #1 SMP Tue Apr 29 13:16:15 EDT 2008 x86_64 x86_64 x86_64 GNU/Linux
(atlsdbal008a)/>dspmqver
Name: WebSphere MQ
Version: 6.0.2.6
CMVC level: p600-206-090217
BuildType: IKAP - (Production)
(atlsdbal008a)/>df -a | grep mqm
/dev/mapper/vgRPRlog01-lvvar_mqm
1032088 35876 943784 4% /var/mqm
/dev/mapper/vgRPRlog01-lvvar_mqm_err
507748 19058 462476 4% /var/mqm/errors
/dev/mapper/vgRPRlog01-lvvar_mqm_log
507748 281980 199554 59% /var/mqm/log
(atlsdbal008a)/>ls -ltr /var/mqm
total 62
drwx------ 2 mqm mqm 16384 May 6 00:07 lost+found
drwxrwsrwx 2 mqm mqm 4096 May 13 22:07 trace
-rw-rw-r-- 1 mqm mqm 1373 May 13 22:07 service.env
drwxrwsr-x 2 mqm mqm 4096 May 13 22:07 exits64
drwxrwsr-x 2 mqm mqm 4096 May 13 22:07 exits
drwxrwsr-x 3 mqm mqm 4096 May 13 22:07 conv
drwxrwsr-x 2 mqm mqm 4096 May 13 22:07 config
drwxrw---- 2 mqm mqm 4096 May 18 15:08 scripts
drwxr-sr-x 2 mqm mqm 4096 Jun 3 21:59 tivoli
drwxrwsrwx 3 mqm mqm 5120 Jun 28 06:09 errors
-rw-rw-r-- 1 mqm mqm 2176 Jun 29 11:45 mqs.ini
drwxrwsr-x 4 mqm mqm 1024 Jun 29 11:45 log
drwxrwsr-x 4 mqm mqm 4096 Jul 6 12:38 qmgrs
(atlsdbal008a)/>ls -ltr /var/mqm/log
total 13
drwx------ 2 mqm mqm 12288 May 6 00:08 lost+found
drwxrwx--- 3 mqm mqm 1024 May 15 13:53 QMNARPRQ1
(atlsdbal008a)/>ls -ltr /var/mqm/qmgrs
total 8
drwxrwsr-x 2 mqm mqm 4096 May 13 22:07 @SYSTEM
drwxrwsr-x 26 mqm mqm 4096 Jun 21 22:19 QMNARPRQ1 |
In this case /opt/mqm is local to both primary & failover linux systems
| Code: |
| There's simple, and there's dangerous, and I consider what you are doing is dangerous. |
Would you mind sharing a hypothetical scenario where in it could be dangerous
Last edited by sunny_30 on Wed Jul 15, 2009 6:59 am; edited 1 time in total |
|
| Back to top |
|
|
| exerk |
| Posted: Wed Jul 15, 2009 6:58 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
So, no queue managers configured on that system then?
Why dangerous? I want to be there the first time someone, or something, inadvertantly of course, tries to start the queue manager on both nodes...
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| sunny_30 |
| Posted: Wed Jul 15, 2009 7:07 am Post subject: |
|
|
Master
Joined: 03 Oct 2005
Posts: 258
|
Note: I edited my recent post above around the same time you responded, added the output from primary-node.
| Code: |
| Why dangerous? I want to be there the first time someone, or something, inadvertantly of course, tries to start the queue manager on both nodes... |
Thanks for the response.
I can test this for you right away
primary:
| Code: |
(atlsdbal008a)/>dspmq
QMNAME(QMNARPRQ1) STATUS(Running) |
failover:
| Code: |
-bash-3.2$ dspmq
-bash-3.2$ strmqm QMNARPRQ1
AMQ8118: WebSphere MQ queue manager does not exist. |
|
|
| Back to top |
|
|
| bruce2359 |
| Posted: Wed Jul 15, 2009 7:11 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9475
Location: US: west coast, almost. Otherwise, enroute.
|
| Quote: |
| Why dangerous? I want to be there the first time someone, or something, inadvertantly of course, tries to start the queue manager on both nodes... |
Until it is tested and proven to function correctly, a DR (or HA) plan is just that - a plan.
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
|
|
