| Author |
Message
|
| zpat |
 Posted: Mon Mar 04, 2013 7:45 am Post subject: Exactly what does the command mqsiaddbrokerinstance do? |
 |
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
I am testing the set up of WMB 7 on AIX using HA/CMP (not MI). I've created the primary node broker, and now it says to run mqsiaddbrokerinstance to add it to the secondary node.
| Quote: |
1.To create a broker, mount the shared resource onto your primary node and use the following mqsicreatebroker command with the -e parameter to specify your shared resource location.
mqsicreatebroker MyBroker -q MQ1 -e /MQHA/MyBroker/where:
◦MyBroker is the name of the broker.
◦MQ1 is the name of the queue manager.
◦/MQHA/MyBroker/ is the directory for your shared resource.
2.To add another broker instance, mount the shared resource onto your secondary nodes and use the following mqsiaddbrokerinstance command.
mqsiaddbrokerinstance MyBroker –e /MQHA/MyBroker/where:
◦MyBroker is the name of the broker.
◦/MQHA/MyBroker/ is the directory for your shared resource.
|
My question is what exactly does mqsiaddbrokerinstance do in terms of which files or directories is it updating?
Because it's a real hassle to move our HA shared disk over just to define the secondary side, this is not required for MQ (as you can just edit the mqs.ini file) and it was not required for the old hamqsiaddbrokerstandy command either.
So I would like to find a way to achieve the same effect without mounting the HA disks on the secondary node, if at all possible. |
|
| Back to top |
|
 |
| zpat |
| Posted: Tue Mar 05, 2013 1:18 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
| I have now posed this question in a PMR, will report back any useful info. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Tue Mar 05, 2013 2:38 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
more or less the same thing that addmqinf does for WMQ...
i.e. tell the current installation about the shared instance... 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| zpat |
| Posted: Wed Mar 06, 2013 12:47 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
I understand its function, what I don't understand is exactly which file(s) it is updating and why it has to have the HA disks online to do it.
addmqinf - also checks for the presence of the MQ HA filesystem, but I can get around that by simply editing mqs.ini instead of using addmqinf.
So what I am looking for is the same workaround for WMB HA.
I really don't want to bring these disks online to the secondary node just to create a new broker - as it's a major change management hassle to arrange the downtime of the other components in the same HA group. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Mar 07, 2013 5:36 am Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| zpat wrote: |
I understand its function, what I don't understand is exactly which file(s) it is updating and why it has to have the HA disks online to do it.
addmqinf - also checks for the presence of the MQ HA filesystem, but I can get around that by simply editing mqs.ini instead of using addmqinf.
So what I am looking for is the same workaround for WMB HA.
I really don't want to bring these disks online to the secondary node just to create a new broker - as it's a major change management hassle to arrange the downtime of the other components in the same HA group. |
This you have to explain some. I would expect that you create both instances when you create the broker. This does not seem to have been the way you are looking to build it. So please elaborate... what else do you have in the HA group that is already live??
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| mqjeff |
| Posted: Thu Mar 07, 2013 5:45 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
| fjb_saper wrote: |
| zpat wrote: |
I understand its function, what I don't understand is exactly which file(s) it is updating and why it has to have the HA disks online to do it.
addmqinf - also checks for the presence of the MQ HA filesystem, but I can get around that by simply editing mqs.ini instead of using addmqinf.
So what I am looking for is the same workaround for WMB HA.
I really don't want to bring these disks online to the secondary node just to create a new broker - as it's a major change management hassle to arrange the downtime of the other components in the same HA group. |
This you have to explain some. I would expect that you create both instances when you create the broker. This does not seem to have been the way you are looking to build it. So please elaborate... what else do you have in the HA group that is already live?? |
He is trying to create both instances when he creates the broker.
He is trying to do this without doing an HA failover from the primary to the secondary.
Apparently, it is complicated in his environment to do a manual failover. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Mar 07, 2013 6:11 am Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
So how can he guarantee that the secondary will ever work? That the share will attach correctly on fail over to the secondary?
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| zpat |
| Posted: Thu Mar 07, 2013 9:20 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
Yes at some point I will test it, that is not the issue here. I want to separate out the build process from the testing process.
Failover is not hard - agreeing a date/time to take down all the other QMs and brokers in the same HA resource group is very hard. In our case with SRDF there is also the matter of cross-site storage changes. It can take literally months to agree a date because every project using WMB feels they have a veto on any change which might affect them - even when it won't! And of course every project is top priority with tight deadlines! |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Mar 07, 2013 4:26 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
you should set up multiple HA resources and not bundle all your qmgrs in the same HA resource... This should make such a failover a little bit easier.
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| zpat |
| Posted: Fri Mar 08, 2013 3:46 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
Yes, but increasing the current resource group numbers would have even worse implications because the HA service addresses would change and every application that references the QM or the broker would need changing.
Also I would still get projects blocking such a failover of another resource group on the basis that it might affect them, even when it won't unless someone gets it wrong and selects the wrong one etc. |
|
| Back to top |
|
|
| zpat |
| Posted: Thu Mar 14, 2013 1:07 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
OK, I have an answer on this one, although of course this is not the officially supported method.
To replicate the effect of mqsiaddbrokerinstance, all that is needed is to create one directory and one file on the secondary node.
/var/mqsi/registry/<brokername>/HASharedWorkPath
The contents of this file (the path to the shared filesystem) being identical to the same file on the primary node. This allows the HA setup to be performed without failing the storage over. |
|
| Back to top |
|
|
| zpat |
| Posted: Thu May 02, 2013 11:36 pm Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
Having used this method and failed HA over, I can report that it does work (just make sure you don't get a spurious LF character in that file).
Any know how to make VI (a VIle editor) edit a file without adding a line feed (or CR) to it? |
|
| Back to top |
|
|
| Vitor |
| Posted: Fri May 03, 2013 5:52 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| zpat wrote: |
| Any know how to make VI (a VIle editor) edit a file without adding a line feed (or CR) to it? |
Not move the file from Windows? 
Seriously, the only time I get CR characters at the end of a line in vi is when I hit the key. This has been true for a number of years on a number of sites. If you move the file from Windoze and whatever moves the file buggers up the conversion you'll get spurious LF as well (^M) but I've not seen what you're describing.
Note that this is the pure vi editor I speak of, not vile or a clone. My view; anything with any connection to emacs is inherently evil and is probably adding characters out of spite.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| zpat |
| Posted: Fri May 03, 2013 6:01 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
So on AIX, I vi a new file. I press i for insert and type in 123 as three characters, press escape :qw
The file size is 4 and not 3 bytes. This is the problem.
No Windoze required! |
|
| Back to top |
|
|
| mqjeff |
| Posted: Fri May 03, 2013 8:04 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
Have you considered
Have you considered that even on Unix, files may be allocated in blocks rather than in specific character sizes...
have you considered that multi-byte characters may cause a file to show a different byte count than the number of characters in the file...
I'm sure you have considered these things...  |
|
| Back to top |
|
|
|
|
