<div dir="ltr"><div dir="ltr">On Thu, Aug 5, 2021 at 3:23 PM Andreas Dilger <<a href="mailto:adilger@whamcloud.com">adilger@whamcloud.com</a>> wrote:<br></div><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div style="overflow-wrap: break-word;">
On Aug 5, 2021, at 13:29, Nathan Dauchy wrote:<br>
<div>
<blockquote type="cite"><div><div dir="ltr"><div>Andreas, thanks as always for your insight. Comments inline...</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Thu, Aug 5, 2021 at 10:48 AM Andreas Dilger <<a href="mailto:adilger@whamcloud.com" target="_blank">adilger@whamcloud.com</a>> wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div>On Aug 5, 2021, at 09:28, Nathan Dauchy via lustre-discuss <<a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a>> wrote:<br>
<div>
<blockquote type="cite">
<div>
<div dir="ltr">
<div>Question: Is it possible that a flash journal device on an ext4 filesystem can reach a point where there are not enough clean blocks to write to, and they can suffer from very degraded write performance?<br></div></div></div></blockquote></div></div></blockquote></div></div></div></blockquote></div></div></blockquote><div>... </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div style="overflow-wrap: break-word;"><div><blockquote type="cite"><div><div dir="ltr"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><div><blockquote type="cite"><div><div dir="ltr"><div>
</div>
</div>
</div>
</blockquote>
</div></div></blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><div><blockquote type="cite"><div><div dir="ltr"><div>Another related question would be how to benchmark the journal device on it's own, particularly write performance, without losing data on an existing file system; similar to the very useful obdfilter-survey tool, but at a lower level. But I am
primarily looking to understand the nuances of flash devices and ldiskfs external journals a bit better.</div>
</div>
</div>
</blockquote>
</div>
<div>While the external journal device has an ext4 superblock header for identification (UUID/label), and a feature flag that prevents it from being mounted/used directly, it is not really an ext4 filesystem, just a flat "file". You'd need to remove
it from the main ext4/ldiskfs filesystem, reformat it as ext4 and mount locally, and then run benchmarks (e.g. "dd" would best match the JBD2 workload, or fio if you want random IOPS) against it. You could do this before/after trim (could use fstrim at this
point) to see if it affects the performance or not.</div>
</div>
</blockquote>
<div><br>
</div>
<div>OK, thanks for confirming that there is no magic ext4 journal benchmarking tool. I'll stop searching. ;-)</div>
</div>
</div>
</div>
</blockquote>
</div><div>Note that there *are* some journal commit statistics - /proc/fs/jbd2/<dev>/info that you might be able to compare between devices. Probably the most interesting is "average transaction commit time", which is how long it takes to write the blocks to the
journal device after the transaction starts to commit.</div>
<div><div dir="auto" style="color:rgb(0,0,0);letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration:none"><div dir="auto" style="color:rgb(0,0,0);letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration:none"><div dir="auto" style="color:rgb(0,0,0);letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration:none"><div dir="auto" style="color:rgb(0,0,0);letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration:none"><div dir="auto" style="color:rgb(0,0,0);letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration:none"><div dir="auto" style="color:rgb(0,0,0);letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration:none"><div><br></div></div></div></div></div></div></div></div></div></blockquote><div><br></div><div>Oh, that is interesting!</div><div><br></div><div>The "average transaction commit time" seems to fluctuate, possibly with load, and doesn't have an obvious correlation to the slower OSTs. But perhaps I'll look at it again when running a clean benchmark during a future dedicated time.</div><div><br></div><div>I _did_ find a stark difference in other metrics though:</div><div><br></div><div># pdsh -g oss "grep 'handles per' /proc/fs/jbd2/md*/info" | sort<br>lfs4n04: 17 handles per transaction<br>lfs4n05: 18 handles per transaction<br>lfs4n06: 18 handles per transaction<br>lfs4n07: 18 handles per transaction<br>lfs4n08: 17 handles per transaction<br>lfs4n09: 17 handles per transaction<br>lfs4n10: 18 handles per transaction<br>lfs4n11: 18 handles per transaction<br>lfs4n12: 18 handles per transaction<br>lfs4n13: 17 handles per transaction<br>lfs4n16: 192 handles per transaction<br>lfs4n17: 178 handles per transaction<br>lfs4n18: 198 handles per transaction<br>lfs4n19: 192 handles per transaction<br># pdsh -g oss "grep 'logged blocks per' /proc/fs/jbd2/md*/info" | sort<br>lfs4n04: 24 logged blocks per transaction<br>lfs4n05: 24 logged blocks per transaction<br>lfs4n06: 25 logged blocks per transaction<br>lfs4n07: 25 logged blocks per transaction<br>lfs4n08: 24 logged blocks per transaction<br>lfs4n09: 24 logged blocks per transaction<br>lfs4n10: 25 logged blocks per transaction<br>lfs4n11: 24 logged blocks per transaction<br>lfs4n12: 24 logged blocks per transaction<br>lfs4n13: 24 logged blocks per transaction<br>lfs4n16: 103 logged blocks per transaction<br>lfs4n17: 98 logged blocks per transaction<br>lfs4n18: 106 logged blocks per transaction<br>lfs4n19: 103 logged blocks per transaction<br></div><div><br></div><div>The last 4 nodes are the expansion OSTs which are performing better. What does that difference indicate?</div><div><br></div><div>Thanks again,</div><div>Nathan</div><div><br></div></div></div>