large customer upgrade v7 -> v8

[AlexL]

We have a rather large customer that needs to be upgraded from v7.1 to v8.4, about 25k licensed users, 4 buckets each over 300TB in size.
Any specific guidance or considerations?

[Polina]

Hi Alex,

Is it about upgrading an entire service provider environment or an installation/server protecting just one customer?

[AlexL]

a single customer, one organization

[Polina]

@bramdelaat Could you please assist?

[Pieter.V]

I will drop you an email @AlexL and we can take it from there. Thanks!

[AlexL]

thanks Pieter

[kevlahau]

Make sure you have plenty of room on the C:/ drive .. I had 90TB free at start of upgrade but ran out of space on indexing our largest customer repo 135T

[karsten123]

90TB free on C:? really?

[sumeet]

Can we please get an update on how did this upgrade go?

Even we have few customers or deployments of this size or greater that needs to be upgraded. Would help to get more details.

[AlexL]

Upgrade went fairly smooth and faster than expected. Repo size is apparently not a deciding factor, probably the number of object in the repos is much more an issue.

We doubled the cpu and mem on the server (not the proxies), we also set maintenance to monthly (we had it on daily). And we disabled all jobs.

Disk seems to be the primary factor, use the fastest you have during upgrading.

Disk space, you need postgress space for the configdb and for all persistent cache, about 80-100% for configdb and 50-80% of persistent cache size. At least for us that was the case in our upgrades.

After upgrade we waited for the configuration database uograde to complete (see status bar bottom right), after that upgrade the repos, try 1 first, decide later if you want multiple concurrent upgrades, we had 4, did 1 and 2 consecutive, did 3 and 4 concurrent. Don’t freak out if it stays at 98% for a hour or more, eventually it will continue with indexing, indexing is usually 2 to 4 times faster than the repo upgrade.

Do not start any job before all repos are completely upgraded.

Start jobs, one by one or in batches, I would not do all of them together. The first time a job runs after upgrading from v7 to v8 additional actions are taking place apparently. First run might take much longer, especially finishing the job maybtake a very long time, I’ve seen jobs literally takes hours completing. Also disk usage will increase a bit during this fase.

General advice, be patient, let things run.

If you get PostgreSQL transient errors and/or index 0 errors, you might need some time out changes in xml files, best to contact support.

Afterwards, delete persistent caches, configdb backup, enable jobs. And possibly enable Teams chat backup if you hadn’t yet.

Good luck,
Alex

[sumeet]

Thanks Alex - appreciate your inputs.

Did you upgrade to v8.4 or v8.3?
If v8.4 - how are the jobs performing post upgrade - primarily from stability point of view.

For one of our clients, with sharepoint backup, support asked us to upgrade to v8.4 to resolve the issue - but post v8.4 upgrade, issue seems to have got worse.
Just want to know your experience.

[AlexL]

The last two installations where upgraded to 8.4, all others to 8.3 and the first few to 8.2. All are running 8.4 now. We changed all to last version only, that seems to help.
Everything runs pretty fine now, any specific issues you are having?