Page 1 of 2

Alerting the user for disk errors

Posted: Sun Jan 02, 2011 10:46 am
by NeverSimple
Hello,

I just build a new HDA and wanted to move all my data from the old one. Because having two HDA's 'online' did not seem like a good idea, I installed Fedora 14 on the old one, so I could access my data drives. As soon as the install finished and I logged in on the desktop I was greeted with a HUGE warning pop-up telling me that one of my disks had lots of bad sectors and was probably failing. And sure enough, when trying to copy data this became a very real problem. (it was one of 2 drives from the greyhole pool and not for all my files I had activated extra copies)

I have been using the Amahi server without ever noticing anything. I don't know how I should have. It doesn't seem that Amahi is configured for sending mails if it detects problems. There is also nothing on the dashboard for alerting users to problems. On a server that's mainly used for storing files something like that should really be integrated, especially if Fedora already implemented it! On my old WHS I got mails for things like fan speed too low, disk storage lower than a certain threshold, cpu temps, etc.

I would suggest putting something like this high on some 'to-do' list (if it isn't already). Having a lot of apps is all very nice, but if the foundation is breaking down beneath them without anyone noticing ....

Sorry, I'm a bit :( .... If I missed anything or set up my server wrong PLEASE tell me!

Richard

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 10:56 am
by gboudreau
Greyhole sends emails when it notices something wrong, but Amahi is not configured to send emails by default.
It's something you have to do yourself, if you want to receive such warnings.

I myself enable outgoing emails on my server, and wrote the procedure on the wiki: http://wiki.amahi.org/index.php/Enable_Outgoing_Emails

Note that Greyhole doesn't monitor the problem you mention. I'm guessing either Fedora or Amahi does / should be able to send emails for such problems.

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 11:38 am
by NeverSimple
Greyhole sends emails when it notices something wrong, but Amahi is not configured to send emails by default.
It's something you have to do yourself, if you want to receive such warnings.

I myself enable outgoing emails on my server, and wrote the procedure on the wiki: http://wiki.amahi.org/index.php/Enable_Outgoing_Emails
Nice instructions. The part about 'copy-paste' is no joke.... I followed your instructions: Is there any way to trigger an outgoing mail? To test if it's working?
Note that Greyhole doesn't monitor the problem you mention. I'm guessing either Fedora or Amahi does / should be able to send emails for such problems.
I've got no idea how greyhole detects that something is wrong but I used rsync to copy the data from my old server to the new one. I used the logfile option and it told me all the files that had read errors and it gave up on. Maybe greyhole could search such a log for errors and include the results in it's own log? The time it took rsync to try and read those defective files was enormous: it tried several times and could easily be busy for 20 minutes with 1 file, before it gave up and tried the next one...

Richard

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 11:45 am
by gboudreau

Code: Select all

echo test 1 2 | mail -s "Some test message" someone@gmail.com

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 12:53 pm
by NeverSimple
Guillaume,

thanks for your help (and the sendmail voodoo in the wiki).

I'm afraid I'll have to bother you (or someone else) again. I can't send mail. receive the following error:

Code: Select all

553 5.1.8 <root@hda.simpson.home>... Domain of sender address root@hda.simpson.home does not exist
Now I kinda understand what that means. The domain I'm sending from does not exist in any DNS and my ISP blocks that. What I don't know, is what to do about it. Sendmail and I are not exactly on the same level. Or, to put in another way: Huhhh ??? :?

Help?

Richard

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 1:48 pm
by gboudreau
Edit /etc/mail/sendmail.mc: you need a line like this one:

Code: Select all

MASQUERADE_AS(`neversimple.yourhda.com')dnl
As long as the hostname you use is on a valid domain, that should work.

You'll need to

Code: Select all

cd /etc/mail make service sendmail restart
for the change to become active.

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 4:01 pm
by NeverSimple
Looks like you opened up the proverbial 'can of worms'...... :shock:

Believe it or not (I almost don't...) I came up with that solution with the help from my good friends Google en Webmin. But ... it still doesn't work. I've edited through webmin and manually but the problem remains. Something has changed (see below) but the sender is still as it was:

Code: Select all

From MAILER-DAEMON@hda.simpson.home Sun Jan 2 23:38:19 2011 Return-Path: <MAILER-DAEMON@hda.simpson.home> Date: Sun, 2 Jan 2011 23:38:19 +0100 From: Mail Delivery Subsystem <MAILER-DAEMON@neversimple.yourhda.com> To: <root@hda.simpson.home> Content-Type: multipart/report; report-type=delivery-status; boundary="p02McJvL024287.1294007899/hda.simpson.home" Subject: Returned mail: see transcript for details Auto-Submitted: auto-generated (failure) Status: R Part 1: The original message was received at Sun, 2 Jan 2011 23:38:19 +0100 from hda.simpson.home [127.0.0.1] ----- The following addresses had permanent fatal errors ----- <xxxx@xs4all.nl> (reason: 553 5.1.8 <root@hda.simpson.home>... Domain of sender address root@hda.simpson.home does not exist)
The 'From:' part now contains the 'yourhda.com' address, but the 'domain of sender address' is still wrong.

I've generated a dozen or so sendmail.cf's (with Webmin) with options enabled that even looked remotely applicable, but so far i missed the right one or the right combination. :(

Richard

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 4:09 pm
by gboudreau
That tells me your hostname is hda.simpson.home, and that's what you should change.

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 4:24 pm
by NeverSimple
That tells me your hostname is hda.simpson.home, and that's what you should change.
Well, I know. But to be clear: If I follow your previous post to the letter all that get's changed is:

Code: Select all

From: Mail Delivery Subsystem <MAILER-DAEMON@neversimple.yourhda.com>
But the 'domain of sender address' does not, for some reason. I'm sorry.

Thanks for your help, appreciate it. I will see if I can solve this.


Richard

Re: Alerting the user for disk errors

Posted: Sun Jan 02, 2011 4:33 pm
by gboudreau
If you follow my last post to the letter, you're changing sendmail's masquerade_as setting, not your server's hostname.
I think you need to change your server's hostname (in a permanent way; google it if webmin doesn't let you do that), and then reboot. In a terminal hostname should report a valid hostname, which is what I think sendmail is using in the example you pasted above.