Migrate and parallelization

[Pierrot_2]

The migrate support utility is such a time saver. There's one thing that might be up for improvement, it's the "

/opt/opsware/support/bin/migrate -i devlist.pkl.gz --device_list devicelist "

It is a little slow and we don't know where it's at in its processing (number of servers processed).

So I was wondering if we could import like this:

1- split the list used to export (the group members) on import in 5; For example if we do batches of 500 servers. So we would run 5 process on 5 slices with the 1/5 of the device list as argument for each... Would that buy us anything? right now 500 servers takes 6+ hours to import for us.

Thanks

[bengland]

I suspect the large import time is due to the fact that you are taking ALL the device history.   Have you changed the default DAYS_CHANGE_LOG of 180 (days) to something much larger in your system?

# sql "select VALUE from audit_params where name = 'DAYS_CHANGE_LOG'"

Query #1 on Facility_id 1 (C1):
VALUE
-----
180

If this is the case on export you can tell the migrate tool that a small amount of device change history should be taken.  This will greatly speed up import times, importing and exporting this can be quite intensive.

--change_log_days <days>
The number of days from today into the past to consider device
change log data for exporting. Default all data. 0 days will
disable exporting device change log history.

Making any other change as you suggest is too risky and could destabilise the code.

Brett

[Pierrot_2]

Thank you, thank you!

We're migrating another 500 tonight so we'll test . Last night it took 9 hrs for 526 devices.

Pierre