Monitoring RAID status
Hi,
I once extended the BigBrothers Disk Script to show a status dot and the sync status of mirrored disks in front of every filesystems status line. Now I migrated to Hobbit. Nearly all works fine but this extension causes the disk graphs to fail. If I understand it right, the do_disk.c program simply reads each line of the status file and displays the fifth column as the disk usage. Since my outfile from the client has two columns in front of each line this fails. The disk usage is now on column seven.
I can now alter the C code but that's bad if I ever install a new version of Hobbit. So is there any idea out there how to add data to an output file without disturbing the Hobbit's standard functions.
BTW: the disk test on a RedHat Box shows one "graph" to much. It's only a non referenced link, so I think Hobbit tries to interpret things that it cannot read, maybe the title line of the DF commmand's output.
Thanks for help Thorsten Erdmann
Hi Thorsten,
On Thu, Apr 13, 2006 at 03:52:33PM +0200, Thorsten.Erdmann at daimlerchrysler.com wrote:
I once extended the BigBrothers Disk Script to show a status dot and the sync status of mirrored disks in front of every filesystems status line. Now I migrated to Hobbit. Nearly all works fine but this extension causes the disk graphs to fail. If I understand it right, the do_disk.c program simply reads each line of the status file and displays the fifth column as the disk usage. Since my outfile from the client has two columns in front of each line this fails. The disk usage is now on column seven.
You're reading the code correctly. It's not the optimal way of doing it, but it works in most cases.
I can now alter the C code but that's bad if I ever install a new version of Hobbit. So is there any idea out there how to add data to an output file without disturbing the Hobbit's standard functions.
The optimal solution would be to change the code to look for the column header indicating where the used percentage is, so that it would adjust to changes in the column layout automatically. However, this is tricky to get right with the variations in df output that Hobbit must handle.
I don't have an easy solution for you right now, I'm afraid. Other than changing the do_disk code yourself.
BTW: the disk test on a RedHat Box shows one "graph" to much. It's only a non referenced link, so I think Hobbit tries to interpret things that it cannot read, maybe the title line of the DF commmand's output.
Getting that count of filesystems right is another very difficult excercise.
Henrik
I'd recommend simply making a new test + script + column name - like raidstat or something. That way you still get your hobbit graphs, you still check your disks, and it's compatible between upgrades. This way it's a logical separation too - disk (space) status and raid status. If you have a heterogenous network, you could make the extra script "smart" too, and have it distinguish between devices like I do - one for Solaris Solstice software mirror, one for IBM ipssend and one for mdstat.
Just my 2 cents.
On 5/29/06, Henrik Stoerner <henrik at hswn.dk> wrote:
Hi Thorsten,
On Thu, Apr 13, 2006 at 03:52:33PM +0200, Thorsten.Erdmann at daimlerchrysler.com wrote:
I once extended the BigBrothers Disk Script to show a status dot and the sync status of mirrored disks in front of every filesystems status line. Now I migrated to Hobbit. Nearly all works fine but this extension causes the disk graphs to fail. If I understand it right, the do_disk.c program simply reads each line of the status file and displays the fifth column as the disk usage. Since my outfile from the client has two columns in front of each line this fails. The disk usage is now on column seven.
You're reading the code correctly. It's not the optimal way of doing it, but it works in most cases.
I can now alter the C code but that's bad if I ever install a new version of Hobbit. So is there any idea out there how to add data to an output file without disturbing the Hobbit's standard functions.
The optimal solution would be to change the code to look for the column header indicating where the used percentage is, so that it would adjust to changes in the column layout automatically. However, this is tricky to get right with the variations in df output that Hobbit must handle.
I don't have an easy solution for you right now, I'm afraid. Other than changing the do_disk code yourself.
BTW: the disk test on a RedHat Box shows one "graph" to much. It's only a non referenced link, so I think Hobbit tries to interpret things that it cannot read, maybe the title line of the DF commmand's output.
Getting that count of filesystems right is another very difficult excercise.
Henrik
To unsubscribe from the hobbit list, send an e-mail to hobbit-unsubscribe at hswn.dk
On 5/29/06, Henrik Stoerner <henrik at hswn.dk> wrote:
The optimal solution would be to change the code to look for the column header indicating where the used percentage is, so that it would adjust to changes in the column layout automatically. However, this is tricky to get right with the variations in df output that Hobbit must handle.
IMO the percentage used is, these days, less useful. On a 3 TB file system, the difference between 99% and 100% isn't as useful as knowing there are still 5 GB of storage free (or in my particular case knowing there is at least 800 MB free).
Is there any chance of making future versions of hobbit support reporting by the remaining free space, rather than the percentage free?
-- Please keep list traffic on the list.
Rob MacGregor Whoever fights monsters should see to it that in the process he doesn't become a monster. Friedrich Nietzsche
On Tue, May 30, 2006 at 12:13:36PM +0100, Rob MacGregor wrote:
Is there any chance of making future versions of hobbit support reporting by the remaining free space, rather than the percentage free?
Yes, it's one of the two outstanding items on my list before the 4.2 release (the other is the backend-handling of IRIX clients).
Regards, Henrik
participants (4)
-
bigdan＠gmail.com
-
henrik＠hswn.dk
-
rob.macgregor＠gmail.com
-
Thorsten.Erdmann＠daimlerchrysler.com