We recently had a situation where the unix host attached to a clarion array went south before an organized shut-down. Data was subsequently lost due to the fact that the queues were not successfully "flushed" prior to the unix crash. The data that we were missing went back to an hour prior to the crash and it still could not be recovered. I thought a Clarion array protected you from host crashes. How can we be more fault tolerant with this configuration?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Clarion Array crashed
- Thread starter wsbjr
- Start date
I'm not clear... did the clariion crash or the server?
Keep in mind that the clariion array will protect all data written to the disks (even in cache), BUT if your server never sent the data to protect it, how the array can do it? clariion, as others, will protect you from hardware failures but if the corruption comes from the server (i mean, the server sent the data corrupted) clarrion con not protect you from that.
Now, If the problem was the server crash, then you need cluster or somthing like that.
In addition, your problem is quite rare... are you using Oracle? what is your application?
Keep in mind that the clariion array will protect all data written to the disks (even in cache), BUT if your server never sent the data to protect it, how the array can do it? clariion, as others, will protect you from hardware failures but if the corruption comes from the server (i mean, the server sent the data corrupted) clarrion con not protect you from that.
Now, If the problem was the server crash, then you need cluster or somthing like that.
In addition, your problem is quite rare... are you using Oracle? what is your application?
- Thread starter
- #3
We did an export of the database and save the export file on the clarion. An hour later the server was brought down and crashed on the way down HARD. According to EMC, the data was never flushed out of cash and may be on the clarion somewhere but the inode on the host was missing and therefore the data unreachable - does this make sense?
yes, it does... when the server writes to disks, actually the server makes the write to its internal cache (server cache, configured at OS level). So after a while, the cache is flushed to the clariion array, it means to the clariion cache memory and finally to the disk drives. When data is in the clariion cache you are protected, but if not all data was flushed *from* the server cache, then not all data was written to the disks.
I guess, according to your post, that the full export was written to the clariion, but the problem is the filesystem's inodes the problem. What is your unix server? what kind of file system you have? (vxfs, ufs, hfs, etc.) hfs, for example is not very secure (hp-ux) and the server must be configured as *synchronic writes*, vxfs is mor secure.. and so on, each filesystems types have their pros and cons.
I guess, according to your post, that the full export was written to the clariion, but the problem is the filesystem's inodes the problem. What is your unix server? what kind of file system you have? (vxfs, ufs, hfs, etc.) hfs, for example is not very secure (hp-ux) and the server must be configured as *synchronic writes*, vxfs is mor secure.. and so on, each filesystems types have their pros and cons.
- Status
