<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
<meta name="Generator" content="Microsoft Word 15 (filtered medium)">
<!--[if !mso]><style>v\:* {behavior:url(#default#VML);}
o\:* {behavior:url(#default#VML);}
w\:* {behavior:url(#default#VML);}
.shape {behavior:url(#default#VML);}
</style><![endif]--><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
{font-family:Consolas;
panose-1:2 11 6 9 2 2 4 3 2 4;}
@font-face
{font-family:"Segoe UI";
panose-1:2 11 5 2 4 2 4 2 2 3;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
pre
{mso-style-priority:99;
mso-style-link:"HTML Preformatted Char";
margin:0in;
font-size:10.0pt;
font-family:"Courier New";}
span.HTMLPreformattedChar
{mso-style-name:"HTML Preformatted Char";
mso-style-priority:99;
mso-style-link:"HTML Preformatted";
font-family:Consolas;}
span.EmailStyle22
{mso-style-type:personal-reply;
font-family:"Calibri",sans-serif;
color:windowtext;}
.MsoChpDefault
{mso-style-type:export-only;
font-size:10.0pt;}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]-->
</head>
<body lang="EN-US" link="blue" vlink="purple" style="word-wrap:break-word">
<div class="WordSection1">
<p class="MsoNormal">Thanks Raj – I’ve checked all of the nodes in the cluster and they all have peer_credits set to 8, and credits are set to 256. AFAIK that’s quite low – 8 concurrent sends to any given peer at a time. Since I only have two OSSes, for this
client, that’s only 16 concurrent sends at a given moment. IDK if at this level this devolves to the maximum RPC size of 1MB or the current max BRW I have set of 4MB, but in either case these are small MB values.<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">I’ve reached out to Andreas and Patrick to try to get a JIRA account to open a bug, but have not heard back yet. If somebody on-list is more appropriate to assist with this, please ping me. I collected quite a bit of logs/traces yesterday
and have sysrq stacks to share when I can get access to the whamcloud JIRA.<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">Best,<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">ellis<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<div>
<div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in 0in 0in">
<p class="MsoNormal"><b>From:</b> Raj <rajgautam@gmail.com> <br>
<b>Sent:</b> Thursday, January 20, 2022 8:14 AM<br>
<b>To:</b> Patrick Farrell <pfarrell@ddn.com><br>
<b>Cc:</b> Andreas Dilger <adilger@whamcloud.com>; Ellis Wilson <elliswilson@microsoft.com>; lustre-discuss@lists.lustre.org<br>
<b>Subject:</b> [EXTERNAL] Re: [lustre-discuss] Lustre Client Lockup Under Buffered I/O (2.14/2.15)<o:p></o:p></p>
</div>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
<table class="MsoNormalTable" border="0" cellspacing="0" cellpadding="0" align="left" width="100%" style="width:100.0%">
<tbody>
<tr>
<td style="background:#A6A6A6;padding:5.25pt 1.5pt 5.25pt 1.5pt"></td>
<td width="100%" style="width:100.0%;background:#EAEAEA;padding:5.25pt 3.75pt 5.25pt 11.25pt">
<div>
<p class="MsoNormal" style="mso-element:frame;mso-element-frame-hspace:2.25pt;mso-element-wrap:around;mso-element-anchor-vertical:paragraph;mso-element-anchor-horizontal:column;mso-height-rule:exactly">
<span style="font-size:9.0pt;font-family:"Segoe UI",sans-serif;color:#212121">You don't often get email from
<a href="mailto:rajgautam@gmail.com">rajgautam@gmail.com</a>. <a href="http://aka.ms/LearnAboutSenderIdentification">
Learn why this is important</a><o:p></o:p></span></p>
</div>
</td>
<td width="75" style="width:56.25pt;background:#EAEAEA;padding:5.25pt 3.75pt 5.25pt 3.75pt;align:left">
</td>
</tr>
</tbody>
</table>
<div>
<div>
<p class="MsoNormal"><span style="color:black">Ellis, I would also check the peer_credit between server and the client. They should be same.</span><o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
<div>
<div>
<p class="MsoNormal">On Wed, Jan 19, 2022 at 9:27 AM Patrick Farrell via lustre-discuss <<a href="mailto:lustre-discuss@lists.lustre.org">lustre-discuss@lists.lustre.org</a>> wrote:<o:p></o:p></p>
</div>
<blockquote style="border:none;border-left:solid #CCCCCC 1.0pt;padding:0in 0in 0in 6.0pt;margin-left:4.8pt;margin-right:0in">
<div>
<div>
<p class="MsoNormal"><span style="font-size:12.0pt;color:black">Ellis,<o:p></o:p></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:12.0pt;color:black"><o:p> </o:p></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:12.0pt;color:black">As you may have guessed, that function just set looks like a node which is doing buffered I/O and thrashing for memory. No particular insight available from the count of functions there.<br>
<br>
<span style="background:white">Would you consider opening a bug report in the Whamcloud JIRA? You should have enough for a good report, here's a few things that would be helpful as well:</span><br>
<br>
It sounds like you can hang the node on demand. If you could collect stack traces with:
<o:p></o:p></span></p>
<pre><span style="font-size:9.0pt;font-family:Consolas;color:black">echo t > /proc/sysrq-trigger<o:p></o:p></span></pre>
<p class="MsoNormal"><span style="font-size:12.0pt;color:black">after creating the hang, that would be useful. (It will print to dmesg.)<br>
<br>
You've also collected debug logs - Could you include, say, the last 100 MiB of that log set? That should be reasonable to attach if compressed.<o:p></o:p></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:12.0pt;color:black"><o:p> </o:p></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:12.0pt;color:black">Regards,<o:p></o:p></span></p>
</div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><span style="font-size:12.0pt;color:black">Patrick<o:p></o:p></span></p>
</div>
<div class="MsoNormal" align="center" style="text-align:center">
<hr size="2" width="98%" align="center">
</div>
<div id="m_-1787521865844447704divRplyFwdMsg">
<p class="MsoNormal"><b><span style="color:black">From:</span></b><span style="color:black"> lustre-discuss <<a href="mailto:lustre-discuss-bounces@lists.lustre.org" target="_blank">lustre-discuss-bounces@lists.lustre.org</a>> on behalf of Ellis Wilson via
lustre-discuss <<a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a>><br>
<b>Sent:</b> Wednesday, January 19, 2022 8:32 AM<br>
<b>To:</b> Andreas Dilger <<a href="mailto:adilger@whamcloud.com" target="_blank">adilger@whamcloud.com</a>><br>
<b>Cc:</b> <a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a> <<a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a>><br>
<b>Subject:</b> Re: [lustre-discuss] Lustre Client Lockup Under Buffered I/O (2.14/2.15)</span>
<o:p></o:p></p>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
</div>
</div>
<div>
<div>
<div>
<p>Hi Andreas,<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>Apologies in advance for the top-post. I’m required to use Outlook for work, and it doesn’t handle in-line or bottom-posting well.<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>Client-side defaults prior to any tuning of mine (this is a very minimal 1-client, 1-MDS/MGS, 2-OSS cluster):<o:p></o:p></p>
<p><br>
~# lctl get_param llite.*.max_cached_mb<o:p></o:p></p>
<p>llite.lustrefs-ffff8d52a9c52800.max_cached_mb=<o:p></o:p></p>
<p>users: 5<o:p></o:p></p>
<p>max_cached_mb: 7748<o:p></o:p></p>
<p>used_mb: 0<o:p></o:p></p>
<p>unused_mb: 7748<o:p></o:p></p>
<p>reclaim_count: 0<o:p></o:p></p>
<p>~# lctl get_param osc.*.max_dirty_mb<o:p></o:p></p>
<p>osc.lustrefs-OST0000-osc-ffff8d52a9c52800.max_dirty_mb=1938<o:p></o:p></p>
<p>osc.lustrefs-OST0001-osc-ffff8d52a9c52800.max_dirty_mb=1938<o:p></o:p></p>
<p>~# lctl get_param osc.*.max_rpcs_in_flight<o:p></o:p></p>
<p>osc.lustrefs-OST0000-osc-ffff8d52a9c52800.max_rpcs_in_flight=8<o:p></o:p></p>
<p>osc.lustrefs-OST0001-osc-ffff8d52a9c52800.max_rpcs_in_flight=8<o:p></o:p></p>
<p>~# lctl get_param osc.*.max_pages_per_rpc<o:p></o:p></p>
<p>osc.lustrefs-OST0000-osc-ffff8d52a9c52800.max_pages_per_rpc=1024<o:p></o:p></p>
<p>osc.lustrefs-OST0001-osc-ffff8d52a9c52800.max_pages_per_rpc=1024<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>Thus far I’ve reduced the following to what I felt were really conservative values for a 16GB RAM machine:<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>~# lctl set_param llite.*.max_cached_mb=1024<o:p></o:p></p>
<p>llite.lustrefs-ffff8d52a9c52800.max_cached_mb=1024<o:p></o:p></p>
<p>~# lctl set_param osc.*.max_dirty_mb=512<o:p></o:p></p>
<p>osc.lustrefs-OST0000-osc-ffff8d52a9c52800.max_dirty_mb=512<o:p></o:p></p>
<p>osc.lustrefs-OST0001-osc-ffff8d52a9c52800.max_dirty_mb=512<o:p></o:p></p>
<p>~# lctl set_param osc.*.max_pages_per_rpc=128<o:p></o:p></p>
<p>osc.lustrefs-OST0000-osc-ffff8d52a9c52800.max_pages_per_rpc=128<o:p></o:p></p>
<p>osc.lustrefs-OST0001-osc-ffff8d52a9c52800.max_pages_per_rpc=128<o:p></o:p></p>
<p>~# lctl set_param osc.*.max_rpcs_in_flight=2<o:p></o:p></p>
<p>osc.lustrefs-OST0000-osc-ffff8d52a9c52800.max_rpcs_in_flight=2<o:p></o:p></p>
<p>osc.lustrefs-OST0001-osc-ffff8d52a9c52800.max_rpcs_in_flight=2<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>This slows down how fast I get to basically OOM from <10 seconds to more like 25 seconds, but the trend is identical.<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>As an example of what I’m seeing on the client, you can see below we start with most free, and then iozone rapidly (within ~10 seconds) causes all memory to be marked used, and that stabilizes at about 140MB free until at some point it stalls for 20 or more
seconds and then some has been synced out:<o:p></o:p></p>
<p><br>
~# dstat --mem<o:p></o:p></p>
<p>------memory-usage-----<o:p></o:p></p>
<p>used free buff cach<o:p></o:p></p>
<p>1029M 13.9G 2756k 215M<o:p></o:p></p>
<p>1028M 13.9G 2756k 215M<o:p></o:p></p>
<p>1028M 13.9G 2756k 215M<o:p></o:p></p>
<p>1088M 13.9G 2756k 215M<o:p></o:p></p>
<p>2550M 11.5G 2764k 1238M<o:p></o:p></p>
<p>3989M 10.1G 2764k 1236M<o:p></o:p></p>
<p>5404M 8881M 2764k 1239M<o:p></o:p></p>
<p>6831M 7453M 2772k 1240M<o:p></o:p></p>
<p>8254M 6033M 2772k 1237M<o:p></o:p></p>
<p>9672M 4613M 2772k 1239M<o:p></o:p></p>
<p>10.6G 3462M 2772k 1240M<o:p></o:p></p>
<p>12.1G 1902M 2772k 1240M<o:p></o:p></p>
<p>13.4G 582M 2772k 1240M<o:p></o:p></p>
<p>13.9G 139M 2488k 1161M<o:p></o:p></p>
<p>13.9G 139M 1528k 1174M<o:p></o:p></p>
<p>13.9G 140M 896k 1175M<o:p></o:p></p>
<p>13.9G 139M 676k 1176M<o:p></o:p></p>
<p>13.9G 142M 528k 1177M<o:p></o:p></p>
<p>13.9G 140M 484k 1188M<o:p></o:p></p>
<p>13.9G 139M 492k 1188M<o:p></o:p></p>
<p>13.9G 139M 488k 1188M<o:p></o:p></p>
<p>13.9G 141M 488k 1186M<o:p></o:p></p>
<p>13.9G 141M 480k 1187M<o:p></o:p></p>
<p>13.9G 139M 492k 1188M<o:p></o:p></p>
<p>13.9G 141M 600k 1188M<o:p></o:p></p>
<p>13.9G 139M 580k 1187M<o:p></o:p></p>
<p>13.9G 140M 536k 1186M<o:p></o:p></p>
<p>13.9G 141M 668k 1186M<o:p></o:p></p>
<p>13.9G 139M 580k 1188M<o:p></o:p></p>
<p>13.9G 140M 568k 1187M<o:p></o:p></p>
<p>12.7G 1299M 2064k 1197M missed 20 ticks <-- client is totally unresponsive during this time<o:p></o:p></p>
<p>11.0G 2972M 5404k 1238M^C<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>Additionally, I’ve messed with sysctl settings. Defaults:<o:p></o:p></p>
<p>vm.dirty_background_bytes = 0<o:p></o:p></p>
<p>vm.dirty_background_ratio = 10<o:p></o:p></p>
<p>vm.dirty_bytes = 0<o:p></o:p></p>
<p>vm.dirty_expire_centisecs = 3000<o:p></o:p></p>
<p>vm.dirty_ratio = 20<o:p></o:p></p>
<p>vm.dirty_writeback_centisecs = 500<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>Revised to conservative values:<o:p></o:p></p>
<p>vm.dirty_background_bytes = 1073741824<o:p></o:p></p>
<p>vm.dirty_background_ratio = 0<o:p></o:p></p>
<p>vm.dirty_bytes = 2147483648<o:p></o:p></p>
<p>vm.dirty_expire_centisecs = 200<o:p></o:p></p>
<p>vm.dirty_ratio = 0<o:p></o:p></p>
<p>vm.dirty_writeback_centisecs = 500<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>No observed improvement.<o:p></o:p></p>
<p> <o:p></o:p></p>
<p style="background:white"><span style="color:black">I’m going to trawl two logs today side-by-side, one with ldiskfs backing the OSTs, and one with zfs backing the OSTs, and see if I can see what the differences are since the zfs-backed version never gave
us this problem. The only other potentially useful thing I can share right now is that when I turned on full debug logging and ran the test until I hit OOM, the following were the most frequently hit functions in the logs (count, descending, is the first
column). This was approximately 30s of logs:</span><o:p></o:p></p>
<p style="background:white"><br>
<span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 205874 cl_page.c:518:cl_vmpage_page())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206587 cl_page.c:545:cl_page_owner_clear())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206673 cl_page.c:551:cl_page_owner_clear())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206748 osc_cache.c:2483:osc_teardown_async_page())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206815 cl_page.c:867:cl_page_delete())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206862 cl_page.c:837:cl_page_delete0())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206878 osc_cache.c:2478:osc_teardown_async_page())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206928 cl_page.c:869:cl_page_delete())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206930 cl_page.c:441:cl_page_state_set0())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 206988 osc_page.c:206:osc_page_delete())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207021 cl_page.c:179:__cl_page_free())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207021 cl_page.c:193:cl_page_free())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207021 cl_page.c:532:cl_vmpage_page())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207024 cl_page.c:210:cl_page_free())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207075 cl_page.c:430:cl_page_state_set0())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207169 osc_cache.c:2505:osc_teardown_async_page())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207175 cl_page.c:475:cl_pagevec_put())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207202 cl_page.c:492:cl_pagevec_put())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207211 cl_page.c:822:cl_page_delete0())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207384 osc_page.c:178:osc_page_delete())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 207422 osc_page.c:177:osc_page_delete())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 413680 cl_page.c:433:cl_page_state_set0())</span><o:p></o:p></p>
<p style="background:white"><span style="font-size:10.5pt;font-family:"Segoe UI",sans-serif;color:black"> 413701 cl_page.c:477:cl_pagevec_put())</span><o:p></o:p></p>
<p> <o:p></o:p></p>
<p>If anybody has any additional suggestions or requests for more info don’t hesitate to ask.<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>Best,<o:p></o:p></p>
<p> <o:p></o:p></p>
<p>ellis<o:p></o:p></p>
<p> <o:p></o:p></p>
<div>
<div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in 0in 0in">
<p><b>From:</b> Andreas Dilger <<a href="mailto:adilger@whamcloud.com" target="_blank">adilger@whamcloud.com</a>>
<br>
<b>Sent:</b> Tuesday, January 18, 2022 9:54 PM<br>
<b>To:</b> Ellis Wilson <<a href="mailto:elliswilson@microsoft.com" target="_blank">elliswilson@microsoft.com</a>><br>
<b>Cc:</b> <a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a><br>
<b>Subject:</b> [EXTERNAL] Re: [lustre-discuss] Lustre Client Lockup Under Buffered I/O (2.14/2.15)<o:p></o:p></p>
</div>
</div>
<p> <o:p></o:p></p>
<table class="MsoNormalTable" border="0" cellspacing="0" cellpadding="0" align="left" width="100%" style="width:100.0%">
<tbody>
<tr>
<td style="background:#A6A6A6;padding:5.25pt 1.5pt 5.25pt 1.5pt"></td>
<td width="100%" style="width:100.0%;background:#EAEAEA;padding:5.25pt 3.75pt 5.25pt 11.25pt">
<div>
<p style="mso-element:frame;mso-element-frame-hspace:2.25pt;mso-element-wrap:around;mso-element-anchor-vertical:paragraph;mso-element-anchor-horizontal:column;mso-height-rule:exactly">
<span style="font-size:9.0pt;font-family:"Segoe UI",sans-serif;color:#212121">You don't often get email from
<a href="mailto:adilger@whamcloud.com" target="_blank">adilger@whamcloud.com</a>.
<a href="http://aka.ms/LearnAboutSenderIdentification" target="_blank">Learn why this is important</a></span><o:p></o:p></p>
</div>
</td>
<td width="75" style="width:56.25pt;background:#EAEAEA;padding:5.25pt 3.75pt 5.25pt 3.75pt">
</td>
</tr>
</tbody>
</table>
<div>
<p><span style="color:white">On Jan 18, 2022, at 13:40, Ellis Wilson via lustre-discuss <</span><a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a>> wrote:<o:p></o:p></p>
<div>
<blockquote style="margin-top:5.0pt;margin-bottom:5.0pt">
<p> <o:p></o:p></p>
<div>
<div>
<p>Recently we've switched from using ZFS to ldiskfs as the backing filesystem to work around some performance issues and I'm finding that when I put the cluster under load (with as little as a single client) I can almost completely lockup the client. SSH
(even existing sessions) stall, iostat, top, etc all freeze for 20 to 200 seconds. This alleviates for small windows and recurs as long as I leave the io-generating process in existence. It reports extremely high CPU and RAM usage, and appears to be consumed
exclusively doing 'system'-tagged work. This is on 2.14.0, but I've reproduced on more or less HOL for master-next. If I do direct-IO, performance is fantastic and I have no such issues regarding CPU/memory pressure.<br>
<br>
Uname: Linux 85df894e-8458-4aa4-b16f-1d47154c0dd2-lclient-a0-g0-vm 5.4.0-1065-azure #68~18.04.1-Ubuntu SMP Fri Dec 3 14:08:44 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux<br>
<br>
I dmesg I see consistent spew on the client about:<br>
[19548.601651] LustreError: 30918:0:(events.c:208:client_bulk_callback()) event type 1, status -5, desc 00000000b69b83b0<br>
[19548.662647] LustreError: 30917:0:(events.c:208:client_bulk_callback()) event type 1, status -5, desc 000000009ef2fc22<br>
[19549.153590] Lustre: lustrefs-OST0000-osc-ffff8d52a9c52800: Connection to lustrefs-OST0000 (at
<a href="mailto:10.1.98.7@tcp" target="_blank">10.1.98.7@tcp</a>) was lost; in progress operations using this service will wait for recovery to complete<br>
[19549.153621] Lustre: 30927:0:(client.c:2282:ptlrpc_expire_one_request()) @@@ Request sent has failed due to network error: [sent 1642535831/real 1642535833] req@0000000002361e2d x1722317313374336/t0(0) o4-><a href="mailto:lustrefs-OST0001-osc-ffff8d52a9c52800@10.1.98.10" target="_blank">lustrefs-OST0001-osc-ffff8d52a9c52800@10.1.98.10</a>@tcp:6/4
lens 488/448 e 0 to 1 dl 1642535883 ref 2 fl Rpc:eXQr/0/ffffffff rc 0/-1 job:''<br>
[19549.153623] Lustre: 30927:0:(client.c:2282:ptlrpc_expire_one_request()) Skipped 4 previous similar messages<br>
<br>
But I actually think this is a symptom of extreme memory pressure causing the client to timeout things, not a cause.<br>
<br>
Testing with obdfilter-survey (local) on the OSS side shows expected performance of the disk subsystem. Testing with lnet_selftest from client to OSS shows expected performance. In neither case do I see the high cpu or memory pressure issues.<br>
<br>
Reducing a variety of lctl tunables that appear to govern memory allowances for Lustre clients does not improve the situation. <o:p></o:p></p>
</div>
</div>
</blockquote>
<div>
<p> <o:p></o:p></p>
</div>
<p>What have you reduced here? llite.*.max_cached_mb, osc.*.max_dirty_mb, osc.*.max_rpcs_in_flight and osc.*.max_pages_per_rpc?<o:p></o:p></p>
</div>
<div>
<p style="margin-bottom:12.0pt"><o:p> </o:p></p>
<blockquote style="margin-top:5.0pt;margin-bottom:5.0pt">
<div>
<div>
<p>By all appearances, the running iozone or even simple dd processes gradually (i.e., over a span of just 10 seconds or so) consumes all 16GB of RAM on the client I'm using. I've generated bcc profile graphs for both on- and off-cpu analysis, and they are
utterly boring -- they basically just reflect rampant calls to shrink_inactive_list resulting from page_cache_alloc in the presence of extreme memory pressure.<o:p></o:p></p>
</div>
</div>
</blockquote>
<p> <o:p></o:p></p>
</div>
<div>
<p>We have seen some issues like this that are being looked at, but this is mostly only seen on smaller VM clients used in testing and not larger production clients. Are you able to test with more RAM on the client? Have you tried with 2.12.8 installed on
the client?<o:p></o:p></p>
</div>
<p> <o:p></o:p></p>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<div>
<p><span style="color:black">Cheers, Andreas</span><o:p></o:p></p>
</div>
<div>
<p><span style="color:black">--</span><o:p></o:p></p>
</div>
<div>
<p><span style="color:black">Andreas Dilger</span><o:p></o:p></p>
</div>
<div>
<p><span style="color:black">Lustre Principal Architect</span><o:p></o:p></p>
</div>
<div>
<p><span style="color:black">Whamcloud</span><o:p></o:p></p>
</div>
<div>
<p><span style="color:black"> </span><o:p></o:p></p>
</div>
<div>
<p><span style="color:black"> </span><o:p></o:p></p>
</div>
<div>
<p><span style="color:black"> </span><o:p></o:p></p>
</div>
</div>
</div>
</div>
</div>
</div>
<p><span style="color:black"> </span><o:p></o:p></p>
</div>
<p style="margin-bottom:12.0pt"> <o:p></o:p></p>
</div>
<p> <o:p></o:p></p>
</div>
</div>
</div>
</div>
<p class="MsoNormal">_______________________________________________<br>
lustre-discuss mailing list<br>
<a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a><br>
<a href="https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Flists.lustre.org%2Flistinfo.cgi%2Flustre-discuss-lustre.org&data=04%7C01%7Celliswilson%40microsoft.com%7C38f0764bd787466d0e1c08d9dc16cfa1%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637782812813349615%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=hio5uJ0QHqAmZMUBV9XnV45Odpug8xAwEQc1VnIxMr4%3D&reserved=0" target="_blank">http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org</a><o:p></o:p></p>
</blockquote>
</div>
</div>
</div>
</div>
</body>
</html>