Hi Pierrick,
I haven't tried it against 3.14.11. The changes involving
alarm status and severity string names may or may not
be required for 3.14.11. The other changes were done
while we were building against 3.14.9 and should be
compatible with 3.14.11.
Regards,
- Bruce
On 01/28/2013 09:35 PM, Pierrick Hanlet wrote:
Thank you Bruce,
I'll probably not get to look at it again until Wednesday, but I'm
very grateful.
I'm building against 3.14.11. Do you think that there might be
significant
changes?
Cheers,
Pierrick
On 01/28/2013 10:22 PM, Bruce Hill wrote:
Hi Pierrick,
I'm attaching a patch from our 64 bit ChannelArchiver code.
I'm not sure it will help your problem, however, as a quick
scan of the diff's looked to me like all the changes were to
compile w/ Base 3.14.12, clean up warnings and improve
error handling.
We also have a different situation, in that we've been using a
64 bit machine for the archive engine from the start.
I'm not sure about the base code I'm comparing against.
I used the most original version I could find which appears to
be
version 2.9.0, but may or may not have some changes made years
ago
by other groups here at SLAC.
I've also downloaded some updates as of May 2012 from
http://epicschanarch.cvs.sourceforge.net/viewvc/epicschanarch/?view=tar
but haven't yet merged them into our code base.
We haven't changed update_indices.pl or IndexTool/main.cpp
as far as I know. They still use the 32 bit index which hasn't
yet been an issue for us. We have had some trouble with
corrupted files, particularly related to unexpected power
outages,
but I think we also had some trouble when copying index
files to another machine. You might try copying an index file
back to your 32 bit machine to see if it still works there.
Regards,
- Bruce Hill
On 01/26/2013 05:24 PM, Kasemir, Kay wrote:
Hi:
If you simply want to serve existing data, there should be no
need to re-build any indices.
But the original software (ArchiveExport, ...) was not
developed on 32 bit, so it might need some 32bit/64bit fixes.
I believe that there are people out there using the archive
sources on 64 bit, but I don't know if they published their
patches that were required to get everything working on 64
bit.
One issue could be that I think some people moved to 64 bit
specifically to update the binary index and data file
structure to support>2GB index files.
That would of course not be compatible with existing 32 bit
data files.
So you'll have to look for somebody who got the old sources
patched for 64 bit, but preserving the original, 32-bit data
structures.
Thanks,
Kay
________________________________________
From: [email protected]
[[email protected]]
On Behalf Of Pierrick Hanlet [[email protected]]
Sent: Saturday, January 26, 2013 1:15 PM
To: James F Ross
Cc: Tech Talk
Subject: Re: copying archived data from 32-bit to 64-bit
machine
Hi James,
Thanks for your response. I'm using the old archiver and what
I would like to do
is to use my new machine to serve the copied/archived data, so
it does appear
that we have/had a similar problem. I don't, at the moment,
need the archive
engine. I did successfully build the installation on the
64-bit machine.
If I'm not mistaken, ArchiveExport does not require the data
server to be running,
but it does require that the index files be correct. If that
is true, then I'm not even
at the point of concern for the server, but would like to
rebuild the indices. I've
tried update_indices.pl, but get errors with the
archiveconfig.pm file
(not a HASH reference ...). Am I correct in deducing that I
need to recreate my
index files? If so, any guidance on doing so would be
welcome; I must be missing
something when I read the manual.
Thanks,
Pierrick
On 01/25/2013 06:35 PM, James F Ross wrote:
Is this the old channel archiver or the CSS based RDB
archiver?
I did the same thing with the channel archiver and it took a
while to resolve. The heart of the trouble is that the cgi
data server (ArchiveDataServer.cgi for the channel archiver)
was unable to run due to a number of issues. The archive
viewer couldn't see the data because ArchiveDataServer.cgi
couldn't start. The final solution for me was to recompile
the archiver from scratch on the 64 bit machine and use the
xml index files and data from the 32 bit machine. I had to
have a lot of help doing this, so I won't be very helpful on
how to do that I'm afraid. However, my goal was to get the
archiver running as well as being able to see the data.
Perhaps as long as you can get the archive data server working
you don't need to do this. I would suggest transferring
ArchiveDataServer.cgi from the old machine to the new machine
with all of its libraries (the 32 bit ones) as a starting
point.
The channel archiver manual may help. It has a section on
testing this server. It is woefully out of date and quite
terse, but it can at least show you where to place the cgi
server and how to test it:
http://www.slac.stanford.edu/grp/lcls/controls/global/sw/epics/extensions/ChannelArchiver/manual.pdf.
If you are using the RDB archiver, I'm afraid I can't help.
Hope that at least gives you a starting place!
James Ross
On Fri, Jan 25, 2013 at 5:19 PM, Pierrick Hanlet<[email protected]<mailto:[email protected]>>
wrote:
Hi,
I copied some archived data from a 32-bit machine to a 64-bit
machine. I'm not able
to read the data. If I do an ArchiveExport I see only a very
small number of entries.
I'm guessing that I need to regenerate the indices, but I've
not been successful in doing
so; perhaps I'm not understanding the manual.
Please help. I've wasted an inordinate amount of time.
Thanks,
Pierrick
--
"Whether you think you can or think you can't, either way, you
are correct" -- Henry Ford
_______________________________________________________________
Pierrick Hanlet
IIT/Fermilab
+1-630-840-5555<tel:%2B1-630-840-5555> (FNAL)
+1-312-567-5745<tel:%2B1-312-567-5745> (IIT)
+1-630-697-8758<tel:%2B1-630-697-8758>/+44-79-48-860-197<tel:%2B44-79-48-860-197>
(US/UK mobile)
--
"Whether you think you can or think you can't, either way, you
are correct" -- Henry Ford
_______________________________________________________________
Pierrick Hanlet
IIT/Fermilab
+1-630-840-5555 (FNAL)
+1-312-567-5745 (IIT)
+1-630-697-8758/+44-79-48-860-197 (US/UK mobile)
--
"Whether you think you can or think you can't, either way, you
are correct" -- Henry Ford
_______________________________________________________________
Pierrick Hanlet
IIT/Fermilab
+1-630-840-5555 (FNAL)
+1-312-567-5745 (IIT)
+1-630-697-8758/+44-79-48-860-197 (US/UK mobile)
--
Bruce Hill
Member Technical Staff
SLAC National Accelerator Lab
2575 Sand Hill Road M/S 10
Menlo Park, CA 94025
|