Bacula-users

Re: [Bacula-users] Corrupted filesystem using iSCSI

2014-07-24 12:53:45
Subject: Re: [Bacula-users] Corrupted filesystem using iSCSI
From: "Fábio R. Medeiros" <fabior AT unicamp DOT br>
To: Kern Sibbald <kern AT sibbald DOT com>, "Bacula-users AT lists.sourceforge DOT net" <bacula-users AT lists.sourceforge DOT net>
Date: Thu, 24 Jul 2014 13:48:08 -0300
Hello Kern,

Thanks for explanations... The backup solution on my environment will be 
based on iSCSI and Bacula, the equipment is already bought.

I believe the problem I had with iSCSI was caused by different OS 
mounting the target and writing to the disk by it's own way. Let me 
myself clear: a Windows machine understand the iSCSI block on a way 
different than a Linux, but both writes on the same disk on different 
ways... I don't know if this could be the problem...

Does a ZFS filesystem on the storage device (just for block checks) 
could be a workaround?

Does anyone have experience with all these?

Thanks again!!

Regards,

Fábio


On 22-07-2014 05:54, Kern Sibbald wrote:
> Hello,
>
> In my opinion, using network mounted disks for backup is a risky
> business.  Bacula is designed to work with direct attached disks and
> tapes, so when you connect them instead by a network, you introduce more
> errors (networks have more faults than direct attached devices), delays
> (latency), and buffering problems.  Latency problems are particularly
> bad for a SCSI device because it is a very chatty protocol.
>
> Bottom line: a lot of people use NFS/CIFS or iSCSI, but I wouldn't bet
> my data on them.
>
> NFS tends to be very unfriendly with Bacula for two reasons:
>
> 1. If the NFS connection drops, Bacula can wait forever (something dumb
> in the implementation) hanging on a system call.
>
> 2. If the NFS partition fills, you may not know about it until the
> device is closed (end of Job) and then it is too late to do much of
> anything.
>
> Best regards,
> Kern
>
> Best regards,
> Kern
>
> On 07/21/2014 07:20 PM, "Fábio R. Medeiros" wrote:
>> Hello folks!
>>
>> I've running Bacula 5.2 in a FreeBSD box. We use an iSCSI NAS as the
>> storage device...
>>
>> Today I tried to make a restore test and FreeBSD says me that the iSCSI
>> target's filesystem (UFS2+j) was corrupted and needed to be repaired. I
>> ran fsck and all gets fine. Is this normal? If so, are there any
>> workarounds (i.e. spool or "run before/after job" directives) in order
>> to minimize the need of these checks?
>>
>> Does anyone have experience with Bacula and iSCSI devices?
>>
>> Here is my Device resource config on bacula-sd.conf:
>>
>> Device {
>>     Name = STORAGE
>>     Media Type = File
>>     Archive Device = /mnt/bkp
>>     LabelMedia = Yes
>>     Random Access = Yes
>>     AutomaticMount = Yes
>>     RemovableMedia = No
>>     AlwaysOpen = Yes
>>     Minimum Block Size = 32768
>>     Maximum Block Size = 32768
>>     Block Checksum = Yes
>> }
>>
>> Regards!!
>>
>> Fábio
>>
>> ------------------------------------------------------------------------------
>> Want fast and easy access to all the code in your enterprise? Index and
>> search up to 200,000 lines of code with a free copy of Black Duck
>> Code Sight - the same software that powers the world's largest code
>> search on Ohloh, the Black Duck Open Hub! Try it now.
>> http://p.sf.net/sfu/bds
>> _______________________________________________
>> Bacula-users mailing list
>> Bacula-users AT lists.sourceforge DOT net
>> https://lists.sourceforge.net/lists/listinfo/bacula-users
>>
> a
>
>

------------------------------------------------------------------------------
Want fast and easy access to all the code in your enterprise? Index and
search up to 200,000 lines of code with a free copy of Black Duck
Code Sight - the same software that powers the world's largest code
search on Ohloh, the Black Duck Open Hub! Try it now.
http://p.sf.net/sfu/bds
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

<Prev in Thread] Current Thread [Next in Thread>