Grouper Users - Open Discussion List

[Chris Hyzer <>]

Let me know if you are able to complete the upgrade... I would run the loader 
even if you aren't provisioning...  I will see about updating the docs about 
it after discussing it at the next dev call and let you know how it does.

Thanks,
Chris

-----Original Message-----
From: Baron Fujimoto 
[mailto:]
 
Sent: Thursday, October 29, 2015 3:42 PM
To: Chris Hyzer
Cc: Grouper Users
Subject: Re: [grouper-users] Upgrade errors 2.2.1 -> 2.2.2

On Thu, Oct 29, 2015 at 06:49:47PM +0000, Chris Hyzer wrote:
> 
>> After patching 2.2.1, downloading, and unpacking the new
>> grouper.apiBinary-2.2.2, I accepted the recommendation to run a script
>> that copies change log temp records to the change log. This ran for hours,
>> and was still running when I left for the day.
>> 
>> When I returned this morning, I was greeted with the error below[*],
>> followed by the advisory that,
>> 
>> =====
>> You need to revert all patches to upgrade
>> 
>> ################ Checking patch grouper_v2_2_1_api_patch_21
>> Patch: grouper_v2_2_1_api_patch_21: was applied on: 2015/08/10 13:41:24"
>> =====
>> 
>> I'm unsure what remedial steps are actually needed now. grouperInstaller
>> looks like it wants to resume from or retry grouper_v2_2_1_api_patch_21.
>> Should I just let it continue, or do I need to "revert *all* patches" as
>> displayed and start over?
>
>Do you run the grouper loader which copies change log temp to change log 
>continuously?  Were there a lot of records unprocessed in grouper change log 
>temp table?  You need to run a grouper loader process for grouper to 
>function correctly.  It is part of the standard upgrade to uninstall all 
>2.2.1 patches to get to 2.2.2.  So let it revert all the 2.2.1 patches, you 
>can answer yes to each one or revert all at once.

Hmm, no we're not running a grouper loader. I wasn't aware of this
requirement. Did I miss it in the documentation somewhere? The wiki
page for Grouper Loader seems to describe it in terms of a provisioning
system, and not something needed for regular/ongoing operation. We've been
provisioning and maintaining all of our groups via the WS.
<https://spaces.internet2.edu/display/Grouper/Grouper+-+Loader>

Where is this maintenance aspect of the change logs described?

>> Is the amount of time being taken to copy the change logs reasonable? I
>> think our db goes down nightly for a backup. If the the copy change log
>> process can't complete before that happens, will it resume from a previous
>> partial completion, or does it need to start all over from the beginning?
>> I'm concerned about the possibility that may never complete given the
>> db's backup downtime constraint.
>
>I don't know, how many records were there?  Yes, it is not one large 
>transaction, it will make progress, and if you stop it and restart it will 
>pickup from where it left off.  Do some record counts while it runs and see 
>it making progress.

Are these the right tables?

select count(*) from grouper_change_log_entry;      ->  419750
select count(*) from grouper_change_log_entry_temp; -> 3988004

So, maybe yikes.

Could this also be the reason why we occasionally see audit log requests
time out? 

Aloha,
-baron
-- 
Baron Fujimoto 
<>
 :: UH Information Technology Services
minutas cantorum, minutas balorum, minutas carboratum desendus pantorum