Disk Statistics show other partition

[opc31199]

I install ZCS 5.04 in /opt partition and the disk usage statistics show other. How I can fix it? Any idea?

Thank's

[uxbod]

Is /opt mounted on a different volume ? Which partition is it showing ?

[opc31199]

/opt is mounted on diferent volume. It is showing /var volume.

Any idea?

[warmbowski]

I have the same problem. I am wondering if zmdisklog is programmed to properly log logical volumes.

I installed ZCS (OSS) 5.04 on CentOS 5. Here are my partition configurations:

sda - /dev/sda1 - /boot - ext3
sda - /dev/VolGroup00/LogVol00 - / - ext3
sda - /dev/VolGroup00/LogVol01 - swap - swap

sdb - /dev/VolGroup01/LogVol00 - /opt - ext3
sdb - /dev/VolGroup01/LogVol01 - /home - ext3

This is the way I partitioned my drives before installing ZCS. The only graph I see is for VG01/LV01 which is mounted as /home. I would like, at the very least, to show VG01/LV00 data in a graph. But, if possible, I would like to configure both of these partitions to show up as Disk Usage graphs.

Is there a place to configure this, or is the data just pulled automatically by zmdisklog?

[warmbowski]

mount command give this:

[root@mail2 ~]# mount -t ext3
/dev/mapper/VolGroup00-LogVol00 on / type ext3 (rw)
/dev/mapper/VolGroup01-LogVol00 on /opt type ext3 (rw)
/dev/sda2 on /boot type ext3 (rw)
/dev/mapper/VolGroup01-LogVol01 on /home type ext3 (rw)

when I looked in zimbra.log, I see data on each of these partitions at 10 min intervals that is accurate:

May 7 04:10:02 mail2 zimbramon[24759]: 24759:info: 2008-05-07 04:10:02, QUEUE: 0 0
May 7 04:10:03 mail2 zimbramon[24764]: 24764:info: 2008-05-07 04:10:02, DISK: mail2.domain.com: dev: /dev/mapper/VolGroup00-LogVol00, mp: /, tot: 14662, avail: 11689
May 7 04:10:03 mail2 zimbramon[24764]: 24764:info: 2008-05-07 04:10:02, DISK: mail2.domain.com: dev: /dev/mapper/VolGroup01-LogVol00, mp: /opt, tot: 116241, avail: 88184
May 7 04:10:03 mail2 zimbramon[24764]: 24764:info: 2008-05-07 04:10:02, DISK: mail2.domain.com: dev: /dev/sda2, mp: /boot, tot: 99, avail: 77
May 7 04:10:04 mail2 zimbramon[24764]: 24764:info: 2008-05-07 04:10:02, DISK: mail2.domain.com: dev: /dev/mapper/VolGroup01-LogVol01, mp: /home, tot: 16536, avail: 15524

I also see queries in logger_myslow.log for each of these partitions.

So maybe the data is being recorded properly, but for some reason not displayed in graphs in the admin interface.

[warmbowski]

I also see 5 rrd files and corresponding gif files (named disk.mail2.domain.com.day.Disk_Usage_#) in /opt/zimbra/logger/db/work, but still only the graph for the /home partition appears on the disk tab of the admin interface.

Hope this help narrow down the problem.

[Klug]

Are all your partitions ext3 ?

Edit after re-read the thread...
warmbowski's are actually

[uxbod]

Okay, after a bit of digging the code to look at is in /opt/zimbra/jetty/webapps/zimbraAdmin/js/zimbraAdmin/statistics/view/ZaServerDiskStatsPage.js and the code

ZaServerDiskStatsPage.prototype.writeImageHtml = function (periodInt) {
    var periodString = "hour";
    var serverName = this._server.name;
    var periodString = this._getPeriodString(periodInt);

    return AjxBuffer.concat("<img  alt='" + ZaMsg.Stats_Unavailable + "' src='/service/statsimg/disk." , serverName ,
                            ".", periodString,".Disk_Usage_0.gif?nodef=1&rand=", Math.random(), "' onload='javascript:ZaServerDiskStatsPage.callMethod(",
                            this.__internalId , ",ZaServerDiskStatsPage.prototype.loadNextImage,[this.parentNode," ,
                            periodInt , ", 0])' onerror='javascript:AjxCore.objectWithId(", this.__internalId ,
                            ").stopLoadingImages(this,0)'><br>");
};

so its starts with image 0.gif and should continue to load the other partitions, until it hits 50 or no image is found. Now time to try and debug

[Klug]

I've checked on my server, I have 4 gifs in the directory.

The first of them are from the previous server (I upgraded the server last month), I don't want it anymore (# 1).

The three others (# 2, 3 and 4) are the one for the "new" server, seem to be correctly computed but do not appear in the stats panel of the Admin UI.

So it looks to me there's two separated issues :
. ZCS should stop computing stats for non existing partitions (and totaly remove the existing stats/graphs?)
. ZCS does not show stats for "new" partitions (new server)

warmbowski, can you check the content of all the GIF files, in order to know if the one that shows up (/home for you) is GIF #1 or another one (that would be different from my own issue then) ?

[uxbod]

Indeed Klug that is correct. I ended up going into mysql and doing

delete from disk_aggregate where device = '/dev/sda1';

I them removed all the .gifs and the .rrds and ran /opt/zimbra/libexec/zmgengraphs. This created the two rrds files as I expected. The rrds/gifs do match to :-

Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/VolGroup00-LogVol00
                       37G  6.6G   29G  19% /
/dev/xvda1             99M   33M   61M  36% /boot
tmpfs                 1.3G     0  1.3G   0% /dev/shm

and in mysql

mysql> select distinct(device) from disk_aggregate;
+---------------------------------+
| device                          |
+---------------------------------+
| /dev/mapper/VolGroup00-LogVol00 | 
| /dev/xvda1                      | 
+---------------------------------+
2 rows in set (0.00 sec)

I am sure it is a bug in the js code as it should loop through all gifs until one is not found. Will ask Mike to take a look as I think it should be raised in Bugzilla.