Unsolved
This post is more than 5 years old
111 Posts
0
4069
June 3rd, 2009 09:00
SRDF adaptive copy and sync modes
Hi guys,
I want to write a script that does the following:
1) change RDF DG to acp_disk (for source performance issues).
2) establish the RDF session
3) wait until ~1% of total tracks is invalid.
4) change to sync mode
5) wait until all devices are synchronized
6) split the RDF session.
I'm wondring what can I do to simplify the 3rd step? Is there an easy way to know that there are low number of invalid tracks? Can the skew option help me here?
Thanks,
Maxim
Edit:
Maybe a 'symrdf verify' command...?
Message was edited by:
MaximUnited
I want to write a script that does the following:
1) change RDF DG to acp_disk (for source performance issues).
2) establish the RDF session
3) wait until ~1% of total tracks is invalid.
4) change to sync mode
5) wait until all devices are synchronized
6) split the RDF session.
I'm wondring what can I do to simplify the 3rd step? Is there an easy way to know that there are low number of invalid tracks? Can the skew option help me here?
Thanks,
Maxim
Edit:
Maybe a 'symrdf verify' command...?
Message was edited by:
MaximUnited
No Events found!
RRR
2 Intern
•
5.7K Posts
1
June 3rd, 2009 13:00
I measured the amount of time a resync usually takes and I simply perform the steps you mentioned based on the clock, not on monitoring the process itself.
I guess that if you do some smart scripting (using *nix "grep" commands) and watch the values go down to 1%, you could do it in a safe way. I must admit I never had the time to develop this script on Windows...
MaximUnited
111 Posts
0
June 3rd, 2009 21:00
The problem with measuring time, that it's not accurate.
You need to assume that daily changes are the same.
RRR
2 Intern
•
5.7K Posts
0
June 4th, 2009 01:00
We monitored the time needed to get in sync for about 2 weeks and we took the worst day and used that as a baseline. Furthermore I logged the process needed to get in sync and stopped SRDF as soon as everything was in sync.
The problem here is, as you already suggested, that as soon as I switch to SRDF A, this could take 10 or 15 minutes instead of only 1 minutes if I switch to SRDF A too fast. I´d monitor and tune the script everyday and after a month or so it was fine.... workable.
I´m sorry I can´t give you the commands as I´m no longer working at that particular DMX site and I don´t remember exactly what the syntax of the scripts were....
MaximUnited
111 Posts
0
June 4th, 2009 01:00
I hope EMC would make our lives easier with something like that built-in. <*cough>HINT<*cough>
Allen Ward
4 Operator
•
2.1K Posts
0
June 8th, 2009 08:00
RRR
2 Intern
•
5.7K Posts
0
June 8th, 2009 09:00
In the upper left corner in Powerlink type "generator" and look for the Symmetrix Procedure Generator. I can see a couple of versions there, so be sure to pick the right one (or all of them). The most recent one is v2.01 from June 5 2009.
Message was edited by:
RRR
You can also visit Home > Support > Product and Diagnostic Tools > Symmetrix Tools > Symmetrix Procedure Generator directly.
MaximUnited
111 Posts
0
June 9th, 2009 01:00
Can't find anything useful there.
Can you post an example script?
RRR
2 Intern
•
5.7K Posts
0
June 9th, 2009 04:00
SET logfile=E:\logs\srdf.log
echo %date% %time% >> %logfile%
echo. >> %logfile%
symrdf -g set mode acp_disk -noprompt
symrdf -g query >> %logfile%
echo symrdf -g resume -noprompt >> %logfile%
symrdf -g resume -noprompt
echo %date% %time% >> %logfile%
echo. >> %logfile%
echo symrdf -g set mode async -noprompt >> %logfile%
symrdf -g set mode async -noprompt
:BEGIN
symrdf -g verify -consistent
IF "%ERRORLEVEL%"=="0" GOTO SUSPEND
echo %date% %time% The symdevs are not in a consistent state yet >> %logfile%
goto BEGIN
:SUSPEND
echo %date% %time% De devices hebben nu een consistente status >> %logfile%
echo symrdf -g suspend -noprompt >> %logfile%
symrdf -g suspend -noprompt
echo %date% %time% >> %logfile%
echo. >> %logfile%
echo ------------------------- >> %logfile%
echo. >> %logfile%
SET logfile=
RRR
2 Intern
•
5.7K Posts
0
June 9th, 2009 04:00
I combined 4 scripts here and each of the 4 were scheduled after one another with the first two right in the beginning and the last 2 some time later, depending on how much deltas need to be replicated.
MaximUnited
111 Posts
0
June 9th, 2009 04:00
But there is no wait time between the 'resume' in ACP state to the async state.
RRR
2 Intern
•
5.7K Posts
0
June 9th, 2009 04:00