When submitting files for classification from the Workplace from a Windows file server with Data Insight client installed,
the files is first copied to \\DataInsight\data\classification\content\ localy on the client then copied to
\\DataInsight\data\classification\content\ on the Classification server.
This behaveur makes the disks on the client to become full and the content fetching safeguard kicks in and stops Winnas and Comm service on the client.
This in turns seems to stop the copying files from the client to the classification servers witch in turn stops the classification.
Everyting seems to run in to a catch 22 status and nothing happens. The Classification server waits for more files but the comunication with the client is lost leaving the client with full disks and no classification...
Is this a way around this behavior?
This is especially problematic with big file servers with milions of files and several TB of data that needs classification.
To circumvent the classification asking for more files than the filer can manage the classification manual suggests that the classification safeguard and winnas safeguard settings be set to the equivalent values.
This article has more details on how to set the classification safeguard. As a reminder the Winnas Safeguard is set under Scanning and Event Monitoring in the Global Settings section of the Settings tab.
Hope that helps Bengt.
I will try this again, configured to stop at 90% and start at 70% and see how it goes...
I have tryed this before and my feeling is when the stop (90%) kicks in the fetching of files stops and in turn the classification seems to stop and never finish. But I try again.
Bengt, the fetching is designed to be in the 1000 objects at a time until the last fetch which is whatever is less than 1000. In some classifications that happens on the first batch. You can contact Veritas support to review your log data interactively to provide other options and explain it within your environment as these are general guidelines and design parameters.
Factors such as disk size, resource availability, network bandwith definitely do play a role in how well classification functions and performance but it seems you are saying the job gets disconnected and never completes.
I have kept the classification running over the weekend and now on monday it has stopped. The disk on the client is 90% full and safegarding has kicked in correctly. On the client the winnas and Comm services has stopped. The classification request has stopped processing data and it seems that the classification server just waits for more files to classify. It manage to classify 259854 files out of 707063. Now we are at the "catch 22 problem"...
Interesting, when I stopped the Watchdog service and manually started the Comm and Winnas service the classification started again... So with Watchdog service stopped everyting works, of course there is other problems with this... :)
Bengt when you disable the watchdog you turn off safeguard so it is logical you could start the services it is designed to stop to prevent filling the drive. Alternatives to remedy the situation is to add resources such as processing power and disk space or instead time in the form of waiting for the classification to complete.
Likely you next issue will be a drive which is full and services not quiesced due to the watchdog no longer monitoring the usage of resources.