Dual method server and BPM workflow methods

We've been running content server 6.5 sp3 patch 09 for a few months. In April we migrated the content server to new load-balanced servers (Windows 2008 R2 x64). Our database is SQL Server 2008 R2. Load balancing our content servers means our method servers are also load balanced. In general, everything is working fine, but one thing. In one of our docbases, we have a BPM workflow with automatic workflow methods. When those BPM automatic workflow methods run, they seem to be getting executed twice (simultaneously) by each method server. So the automatic activity fails (and pauses) with errors that reflect a conflict. Have we missed something in our configuration to avoid method server conflicts or should we only be running one method server?

Find more posts tagged with

Documentum

Comments

Freejin

Hello,

we have been faced with similar problem on 6.6.0.055 P0900 Reporisory and Method Server.

Our bpm automatic activities became run twice.

How to avoid this?

Thanks in advance!

PanfilovAB

If you are on UNIX environment what is the result of the following command?

strings path_to_documentum_binary | grep 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems'

Have you tried to disable WFA task notification feature (wf_agent_notify_newtask=F in dm_server_config)?

Freejin

Yes, we are using UNIX environment.

1. Command (strings path_to_documentum_binary | grep 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems') didn't show and result for 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems'

2. Have you tried to disable WFA task notification feature (wf_agent_notify_newtask=F in dm_server_config)?

Yes, It didn't help us

PanfilovAB


Denis wrote:

Yes, we are using UNIX environment.
1. Command (strings path_to_documentum_binary | grep 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems') didn't show and result for 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems'

Are you sure? For my environment (Linux/Oracle):


strings documentum | grep 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems\|6.6.0'
Agent master acquiring lock before batch update
6.6.0.041

strings documentum | grep 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems\|6.6.0'
Agent master acquiring lock before batch update
6.6.0.055 P3600

i.e. both CS6.6 and CS66P36 contains string "Agent master acquiring lock before batch update". Similar behavior is known for me as CS-39117 (When there is workload from multi server docbase, VERSION_MISMATCH error occurs.) in D6.7SP1, the reason was EMC incorrectly removed database lock in workflow agent algorithm and the difference between correct binary and incorrect is that correct one contains strings like 'Agent master acquiring lock before batch update' and 'Agent master failed to acquire lock before updating workitems', for example:


strings documentum | grep 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems\|6.7.1'
6.7.1020.0067

strings documentum | grep 'Agent master acquiring lock\|Agent master failed to acquire lock before updating workitems\|6.7.1'
Agent master acquiring lock
Agent master failed to acquire lock before updating workitems
6.7.1090.0170


Yes, It didn't help us

You have to options:

Disable workflow agent on second server and consult support
Disable workflow agent on second server and dot not consult support