RE: [U2] Triggers on a NFS file?
-Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Martin Phillips Sent: Friday, October 24, 2008 1:18 AM To: u2-users@listserver.u2ug.org Subject: Re: [U2] Triggers on a NFS file? Hi George, If I have a NFS mounted directory (on a third non-UV system) which houses a UV format file. I have (2) linux system that have that directory mounted, which each have UV that have a VOC pointer to that file. 1F , 2 /mnt/directory/filename , 3 /mnt/directory/D_filename Both UV systems can read/write to this file, but only one system will be writing to the file, the other will only be reading.I realize locking is not respected from one system to another. Be careful! I have seen (and repaired) files that have been damaged in broadly similar situation. As you say, the locking system will not work as the file is being accessed by two completely independent UV systems. The problem that you may have is not with record locking but with group locking that protects the internal structure of the file. Even with only one system writing, it is possible (likely?) that there will be times when the system doing the reading will access a group at the same moment that the other system is modifying an overflow chain. We run into similar timing issues with trading partners when exchanging data via ftp, although this doesn't involve UV hashed files. The issue is similar, though, in that one side needs to know for sure that the other has finished writing before it begins reading to avoid data loss. One way we handle this is by having the writing machine write a file with the same name and a .done extension appended to the end when it is done updating. The writing machine will not attempt to do any further writing to that file as long as the .done file exists. The reading machine will wait until it sees the .done file before it begins reading, and will delete the .done file when it's finished. I think that might also work in this scenario. -John --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Triggers on a NFS file?
Hi George, If I have a NFS mounted directory (on a third non-UV system) which houses a UV format file. I have (2) linux system that have that directory mounted, which each have UV that have a VOC pointer to that file. 1F , 2 /mnt/directory/filename , 3 /mnt/directory/D_filename Both UV systems can read/write to this file, but only one system will be writing to the file, the other will only be reading.I realize locking is not respected from one system to another. Be careful! I have seen (and repaired) files that have been damaged in broadly similar situation. As you say, the locking system will not work as the file is being accessed by two completely independent UV systems. The problem that you may have is not with record locking but with group locking that protects the internal structure of the file. Even with only one system writing, it is possible (likely?) that there will be times when the system doing the reading will access a group at the same moment that the other system is modifying an overflow chain. If I defined a trigger on this file on the system that will only be doing the reading, will it detect any changes to the file when the other UV makes changes? No. Triggers are fired somewhere deep in the UV file system. There is no way in which one system can know about changes applied by the other system. Even with uv/net to avoid the locking problems, the triggers are not going to get run on both systems for an update. Basically System1 will be making changes to that file. And System2 will need to update on it's system based on those changes. Sounds like you need to investigate data replication, changing the network structure if necessary. Martin Phillips Ladybridge Systems Ltd 17b Coldstream Lane, Hardingstone, Northampton, NN4 6DB +44-(0)1604-709200 --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
RE: [U2] Triggers on a NFS file?
The NFS mount is actually set to ro (read only), so I wouldn't think (not to be read as it couldn't happen) the files would get corrupted. Stranger UV things have happened. The file itself is not on a UV system. What I did was a CREATE-FILE then copied the file to the shared directory. Then I did a DELETE-FILE so it was no longer in the original UV's systems linkages. Now the only reference is via a VOC remote pointer. I will have to play around with this (Warnings taken, not lightly). Thanks for the info. George -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Martin Phillips Sent: Friday, October 24, 2008 5:05 AM To: u2-users@listserver.u2ug.org Subject: Re: [U2] RE: Triggers on a NFS file? UV introduced the ALLOWNFS configuration option long ago with a default setting of 0 (no). When I deliver UV training, I take great care to emphasise that setting this to 1 (yes) also means it's all my fault if the file gets corrupted. Martin Phillips --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
[U2] Triggers on a NFS file?
If I have a NFS mounted directory (on a third non-UV system) which houses a UV format file. I have (2) linux system that have that directory mounted, which each have UV that have a VOC pointer to that file. 1F , 2 /mnt/directory/filename , 3 /mnt/directory/D_filename Both UV systems can read/write to this file, but only one system will be writing to the file, the other will only be reading.I realize locking is not respected from one system to another. If I defined a trigger on this file on the system that will only be doing the reading, will it detect any changes to the file when the other UV makes changes? Basically System1 will be making changes to that file. And System2 will need to update on it's system based on those changes. The reason I'm using NFS is that the two UV systems can't interact easily because they are on different networks, but both have access to the system which is exporting the UV formatted file. George Gallen Senior Programmer/Analyst Accounting/Data Division [EMAIL PROTECTED] ph:856.848.9005 Ext 220 The Wyanoke Group http://www.wyanokegroup.com --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
RE: [U2] Triggers on a NFS file?
We run on Unidata 7.x on both Windows, HP/UX and AIX We use FX pointers and NFA to read and write to files on two physically separate systems. Nicholas M Gettino | Director of Development | EnRoute Emergency Systems, an Infor company | office: 813-207-6998 | fax: 678-393-5389 [EMAIL PROTECTED] | www.enroute911.com -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of George Gallen Sent: Thursday, October 23, 2008 5:05 PM To: u2-users@listserver.u2ug.org Subject: [U2] Triggers on a NFS file? If I have a NFS mounted directory (on a third non-UV system) which houses a UV format file. I have (2) linux system that have that directory mounted, which each have UV that have a VOC pointer to that file. 1F , 2 /mnt/directory/filename , 3 /mnt/directory/D_filename Both UV systems can read/write to this file, but only one system will be writing to the file, the other will only be reading.I realize locking is not respected from one system to another. If I defined a trigger on this file on the system that will only be doing the reading, will it detect any changes to the file when the other UV makes changes? Basically System1 will be making changes to that file. And System2 will need to update on it's system based on those changes. The reason I'm using NFS is that the two UV systems can't interact easily because they are on different networks, but both have access to the system which is exporting the UV formatted file. George Gallen Senior Programmer/Analyst Accounting/Data Division [EMAIL PROTECTED] ph:856.848.9005 Ext 220 The Wyanoke Group http://www.wyanokegroup.com --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/