nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Clustering for Teamsite

bpg

Good afternoon,

Can anyone own up to having a Windows-based clustering solution for Teamsite 6.5 running on Veritas cluster service? And if so, are there any caveats to be aware of?

Essentially, would the Teamsite app react gracefully to real-time replication and would the source and replicated backing-stores maintain integrity? We have had all kinds of warnings from Interwoven that this would lead to irreversible corruption of the backing store and that a ton of customisation would be needed to achieve this but I got the impression that they were just trying to frighten us so that we'd buy their consultancy services and make more money....

Any input would be greatly appreciated

Regards,

Brendan
Rogers IT, Rogers Inc

Edited by bpg on 09/09/05 09:37 AM (server time).

Find more posts tagged with

Comments

Migrateduser

If you mean you're using the store when you are changing it
(ie real time modifications underneith TeamSite) that is a bad idea and
the usual bad stuff is going to happen: crashes, corruption, etc.

Jason Heirtzler
Interwoven Engineering

chuckles

We got the same speech after several meetings, but no offer of consulting services either. Just a vague assurance that it may be supported by some time in '07. I think it really is unsupported and unsafe. If you ever get it working, I think that a lot of people here would love to hear about it.

bpg

do you think that, if technically possible in the replicating software, it would be possible to "store up" the changes in teamsite, freeze the app temporarily and then run an instance of the replication so that the source and target volume are updated/synched intermittently?

in this way, changes are not being made in real-time but are being shipped across periodically while the app itself is in a frozen state.

please feel free to let me know if im talking out of my behind, or if this would indeed be achievable

cheers,

brendan
teamsite sys admin
rogers IT, rogers inc.

jed

You are on the right track. This is the only safe way to do it....
1. Freeze backing store
2. Backup/replicate/whatever
3. unfreeze backingstore

Obviously, step #2 may take quite awhile, so something like this is usually preferred:

Preferred: Mount TS backing store on some sort of mirrored device
1. Freeze backing store (few seconds, usually)
2. Break the mirror and take one slice offline (depends on hardware, but a few seconds for us)
3. Unfreeze backing store (users are now live on the system again)
4. replicate/backup/whatever offline slice
5. Introduce the offline slice to the array and let the mirror software sync things up (re-sync depends on hardware, but should happen in the background)

Now, if TS were to crash hard, you could use the replicated/offline copy of the backing store.

In our case, steps #1 and #2 take a few seconds, but when doing an iwfreeze, be aware that an iwfreeze is **not always instantaneous**. If iwserver needs to write a lot out to disk, this may take awhile which may lead to readonly errors to end-users and other errors when CLT's such as iwcallback are run. It would be nice if iwserver queued all of these requests instead of failing or reporting a R/O state. I would love to do an offline backing store sync every 60 minutes, but there is too much risk since users may be presented with errors.

--
Jed Michnowicz
jedm@sun.com
Content Management Engineering
Sun Microsystems

iwovGraduate

Can you explain what you are trying to achieve by replicating the backing store? Clustering as in load balancing? failover to a secondary TS server?
I really doubt if you will acheive either of them just by replicating the backing store.

gzevin

IWOV has stopped selling the HA TeamSite... since 6.5 I guess.

I took part in a project in Sun Cluster environment. We did manage to do something.. but still, the whole clustered strategy needs to be beefed up by IWOV, I think

Greg Zevin, Ph.D. Comp. Sc.
Independent Interwoven Consultant/Architect
Sydney, AU

Johnny

There is a safe way to mirror a store as you described, but that's because it's a one way operation. Clustering is a two way sync - just how would you expect to do that if IWOV engineering don't even have a way?

John Cuiuli

bpg

I spose the sticking point here is that just about every other app in the world can be clustered/replcated quite safely and happily, apart from Teamsite. According to Interwoven, this is because replication will lead to eventual corruption of the backing store. Ive also noticed that if you even so much as fart anywhere near the backing store, then it will become corrupted.....So damned if you do and damned if you dont. Fine business model there...

In a business dominated by 24/7 requirements, why is it that Interwoven will not support a high-availability architecture based on something like Veritas' Clustering Service, which is probably one of the best-supported and robust solutions around? What i want to achieve is a hot standby solution with a sound backup process that causes minimum impact on the production application, as our web instances that are managed by Teamsite are international and so need to be managed 24/7.

In answer to iwovgraduate's earlier query, what I want to ideally achieve is real-time replication of Teamsite changes to a secondary server based on Veritas' Storage foundation for Windows. This secondary server will then be snapped off intermittently so that an offline backup to tape can happen. I'd then like to turn this into a true clustered environment based on Veritas' Clustering Service so that we can have a near-as-dammit 24/7 solution.

I ma being naive in asking for this??

Best regards,

Brendan
Teamsite Sys Admin
Rogers IT, Rogers Inc

10-03-08 BIRT.zip

jed

I wish more customers were making noise about this, or something very similar. Obviously, I would prefer Sun Cluster support, but I will take anything I can get at this point. Have you filed a feature request for this? If so, please let us know the FR #.

--
Jed Michnowicz
jedm@sun.com
Content Management Engineering
Sun Microsystems