<div dir="ltr"><br><br><div class="gmail_quote">2008/8/25 r2d2 <span dir="ltr"><<a href="mailto:r2d2z4j6y8@gmail.com">r2d2z4j6y8@gmail.com</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div dir="ltr"><br><div class="gmail_quote">[snip me]<br><div><br>I choose creating a new fs as I think the JFS filesystem is definitely corrupted. Most of you are thinking of software troubleshooting, so I will prefer EXT3 for this new fs. It is less efficient for recordings (big files) but more robust, I think.</div>
</div></div></blockquote><div><br>I can't really comment on this, although isn't JFS an IBM thing? I suspect it's fairly robust (it's part of AIX after all, and that's a proper unix *ducks*) I don't know about the code in linux though.<br>
<br>Mind you, I suspect that unless your i/o is aproaching the limits on your system (which if you're recording tv it almost certantly isn't) you'll not notice any issues. (I'm using ext3 for both my recording drives fwiw; if you're interested I can probably dig out the IO stats.) <br>
</div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div dir="ltr"><div class="gmail_quote"><div><br>
So as to be sure it is not a hardware issue, I used badblocks as mentioned by Alex.<br><pre>badblocks -n -s /dev/sdb<br>Checking for bad blocks (non-destructive read-write test)<br>Testing with random pattern: done<br>
</pre>I also used smartctl -t long /dev/sdb which doesn't log any error. So I think the harddisk is healthy. I am going to apply this check on the other harddisk to be sure it is a software troubleshooting.</div></div>
</div></blockquote><div><br>Sounds like it. <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div dir="ltr"><div class="gmail_quote">
<div><br>Shouldn't smartctl be enough ?</div></div></div></blockquote><div><br>Nope. There are many errors that SMART can't detect, or failures that happen so quickly that the drive is dead before it does. <br></div>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div dir="ltr"><div class="gmail_quote"><div><br>
Is there a best way to check harddisk ?<br> </div></div></div></blockquote><div>badblocks sounds like a good way. you could probably do something like:<br><br>dd if=/dev/hd0 of=/dev/null <br><br>and check for read errors too, but the reality is modern harddiscs tend to either last for ages until they wear out or go very quickly. If you've got RAID5 or 1 (it is 1 that's striped isn't it?) you should detect the error before you loose the data.<br>
<br>Ian<br><br></div></div><br></div>