<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=gb2312"><meta name=Generator content="Microsoft Word 12 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:宋体;
panose-1:2 1 6 0 3 1 1 1 1 1;}
@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:"\@宋体";
panose-1:2 1 6 0 3 1 1 1 1 1;}
@font-face
{font-family:Tahoma;
panose-1:2 11 6 4 3 5 4 4 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
margin-bottom:.0001pt;
text-align:justify;
text-justify:inter-ideograph;
font-size:10.5pt;
font-family:"Calibri","sans-serif";}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
p.MsoPlainText, li.MsoPlainText, div.MsoPlainText
{mso-style-priority:99;
mso-style-link:"纯文本 Char";
margin:0cm;
margin-bottom:.0001pt;
font-size:10.5pt;
font-family:"Calibri","sans-serif";}
pre
{mso-style-priority:99;
mso-style-link:"HTML 预设格式 Char";
margin:0cm;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:宋体;}
p.MsoAcetate, li.MsoAcetate, div.MsoAcetate
{mso-style-priority:99;
mso-style-link:"批注框文本 Char";
margin:0cm;
margin-bottom:.0001pt;
text-align:justify;
text-justify:inter-ideograph;
font-size:9.0pt;
font-family:"Calibri","sans-serif";}
span.Char
{mso-style-name:"纯文本 Char";
mso-style-priority:99;
mso-style-link:纯文本;
font-family:"Calibri","sans-serif";}
span.Char0
{mso-style-name:"批注框文本 Char";
mso-style-priority:99;
mso-style-link:批注框文本;
font-family:"Calibri","sans-serif";}
span.EmailStyle21
{mso-style-type:personal-reply;
font-family:"Calibri","sans-serif";
color:#1F497D;}
span.HTMLChar
{mso-style-name:"HTML 预设格式 Char";
mso-style-priority:99;
mso-style-link:"HTML 预设格式";
font-family:宋体;}
.MsoChpDefault
{mso-style-type:export-only;
font-size:10.0pt;}
@page WordSection1
{size:612.0pt 792.0pt;
margin:72.0pt 90.0pt 72.0pt 90.0pt;}
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=ZH-CN link=blue vlink=purple style='text-justify-trim:punctuation'><div class=WordSection1><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>I use perf top and try to see where in the kernel code all this time was being spend. Following is what I was seeing:<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>PerfTop: 2036 irqs/sec kernel:100.0% exact: 0.0% [1000Hz cycles], (target_pid: 5336)<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>----------------------------------------------------------------------------------------------------------<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>samples pcnt function DSO<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>_______ _____ _________________________ _________________<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>11265.00 91.0% _spin_lock_irq [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>355.00 2.9% _spin_lock_irqsave [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>249.00 2.0% compaction_alloc [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>235.00 1.9% compact_zone [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>151.00 1.2% get_pageblock_flags_group [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>32.00 0.3% _cond_resched [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>27.00 0.2% copy_page_c [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>8.00 0.1% _spin_lock [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>6.00 0.0% mem_cgroup_del_lru_list [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>5.00 0.0% __wake_up_bit [kernel.kallsyms]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>Then, I use “perf record -g -p 5336” to capture the percent of kernel call and find “compact_zone” is very busy.<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>[root@bj-nx-cip-w87 ~]# perf report --stdio<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'># Events: 47K cycles<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>#<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'># Overhead Command Shared Object Symbol<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'># ........ ......... ..................... ......................................<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>#<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> 91.51% glusterfs [kernel.kallsyms] [k] _spin_lock_irq<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> |<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> --- _spin_lock_irq<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | <o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> |--99.64%-- compact_zone<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | compact_zone_order<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | try_to_compact_pages<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | __alloc_pages_nodemask<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | alloc_pages_vma<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | do_huge_pmd_anonymous_page<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | handle_mm_fault<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | __do_page_fault<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | do_page_fault<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | page_fault<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | | <o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | |--94.45%-- xdr_callmsg_internal<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | | 0x3829b98860<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | | <o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> | --5.55%-- __memcpy_sse2<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'> --0.36%-- [...]<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>Last, I google “compact_zone” and find the article “<a href="http://structureddata.org/2012/06/18/linux-6-transparent-huge-pages-and-hadoop-workloads/">Linux 6 Transparent Huge Pages and Hadoop Workloads</a>”. The sample issues is occurred in hadoop.<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>THP can be disabled by running the following command:<o:p></o:p></span></p><div style='mso-element:para-border-div;border:solid #DDDDDD 1.0pt;padding:4.0pt 4.0pt 4.0pt 4.0pt;background:#F4F5F7'><p class=MsoNormal align=left style='text-align:left;line-height:14.4pt;background:#F4F5F7;vertical-align:baseline;border:none;padding:0cm'><span lang=EN-US style='font-size:9.0pt;font-family:"Courier New";color:#444444'>echo never > /sys/kernel/mm/redhat_transparent_hugepage/enabled<o:p></o:p></span></p></div><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'>CPU usage and work load is normal.<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='color:#1F497D'><o:p> </o:p></span></p><div><div style='border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0cm 0cm 0cm'><p class=MsoNormal align=left style='text-align:left'><b><span lang=EN-US style='font-size:10.0pt;font-family:"Tahoma","sans-serif"'>From:</span></b><span lang=EN-US style='font-size:10.0pt;font-family:"Tahoma","sans-serif"'> gluster-devel-bounces+gluster=163.com@nongnu.org [mailto:gluster-devel-bounces+gluster=163.com@nongnu.org] <b>On Behalf Of </b>Song<br><b>Sent:</b> Friday, June 07, 2013 1:34 PM<br><b>To:</b> 'Stephan von Krawczynski'; 'Pablo'<br><b>Cc:</b> gluster-devel@nongnu.org<br><b>Subject:</b> Re: [Gluster-devel] [Gluster-users] GlusterFS (3.3.1) - performance issues - large number of LOOKUP calls & high CPU usage<o:p></o:p></span></p></div></div><p class=MsoNormal align=left style='text-align:left'><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>We have met same performance issues when open a file. Sometime, it cost more than 10 seconds that a file open.<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>We add some debug info to locate this problem and test again and again, find it's will cost a few seconds when execute 'xdr_callmsg' function in 'rpc_request_to_xdr'.<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>A typical example of log is as below. From <span style='color:red'>[2013-06-06 13:34:59.471004] </span>to <span style='color:red'>[2013-06-06 13:35:04.890363]</span> , 'xdr_callmsg' function cost more than 5 seconds.<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>[2013-06-06 13:34:59.470991] I [rpc-clnt.c:1175:rpc_clnt_record_build_record] 0-gfs1-client-51: (thread_id is 140257410492160 )add for open_slow rpc_fill_request_end<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US style='color:red'>[2013-06-06 13:34:59.471004] I [xdr-rpcclnt.c:87:rpc_request_to_xdr] 0-rpc: (thread_id is 140257410492160 len = 131072 )add for open_slow xdrmem_create_end<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>[2013-06-06 13:34:59.570044] I [client.c:124:client_submit_request] 0-gfs1-client-86: (thread_id is 140257819739904 )add for open_slow rpc_clnt_submit<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>[2013-06-06 13:34:59.570091] I [rpc-clnt.c:1363:rpc_clnt_submit] 0-gfs1-client-86: (thread_id is 140257819739904 )add for open_slow callid end<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>......<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>[2013-06-06 13:34:59.579865] I [client3_1-fops.c:2235:client3_1_lookup_cbk] 0-gfs1-client-5: (thread_id is 140257819739904)add for open_slow lookup_cbk path=/xmail_dedup/gfs1_000/1FA/1B1<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>[2013-06-06 13:34:59.579917] I [client3_1-fops.c:2235:client3_1_lookup_cbk] 0-gfs1-client-6: (thread_id is 140257819739904)add for open_slow lookup_cbk path=/xmail_dedup/gfs1_000/1FA/1B1<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US style='color:red'>[2013-06-06 13:35:04.890363] I [xdr-rpcclnt.c:92:rpc_request_to_xdr] 0-rpc: (thread_id is 140257410492160 )add for open_slow xdr_callmsg_end<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>[2013-06-06 13:35:04.890366] I [client.c:110:client_submit_request] 0-gfs1-client-44: (thread_id is 140257785079552 )add for open_slow create the xdr payload<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>Native client and use 5 glusterfs in one server. When performance issues appear, the cpu usage is as below:<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>top - 13:45:37 up 57 days, 14:04, 4 users, load average: 6.98, 5.38, 4.67<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Tasks: 712 total, 8 running, 704 sleeping, 0 stopped, 0 zombie<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Cpu(s): 3.2%us, 63.5%sy, 0.0%ni, 31.5%id, 1.4%wa, 0.0%hi, 0.4%si, 0.0%st<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Mem: 65956748k total, 55218008k used, 10738740k free, 3362972k buffers<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Swap: 8388600k total, 41448k used, 8347152k free, 37370840k cached<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>13905 root 20 0 554m 363m 2172 R 244.8 0.6 9:51.01 glusterfs<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>13650 root 20 0 766m 610m 2056 R 184.8 0.9 18:24.37 glusterfs<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>13898 root 20 0 545m 356m 2176 R 179.2 0.6 12:04.87 glusterfs<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>13919 root 20 0 547m 360m 2172 R 111.6 0.6 9:16.89 glusterfs<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>22460 root 20 0 486m 296m 2200 S 100.4 0.5 194:59.10 glusterfs <o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>13878 root 20 0 545m 361m 2176 R 99.7 0.6 8:35.88 glusterfs <o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>-----Original Message-----<br>From: <a href="mailto:gluster-users-bounces@gluster.org">gluster-users-bounces@gluster.org</a> [<a href="mailto:gluster-users-bounces@gluster.org">mailto:gluster-users-bounces@gluster.org</a>] On Behalf Of Stephan von Krawczynski<br>Sent: Thursday, June 06, 2013 10:07 PM<br>To: Pablo<br>Cc: <a href="mailto:gluster-users@gluster.org">gluster-users@gluster.org</a><br>Subject: Re: [Gluster-users] GlusterFS (3.3.1) - performance issues - large number of LOOKUP calls & high CPU usage<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>On Thu, 06 Jun 2013 10:39:21 -0300<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Pablo <<a href="mailto:paa.listas@gmail.com"><span style='color:windowtext;text-decoration:none'>paa.listas@gmail.com</span></a>> wrote:<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>> I have never try this (In fact I'm just learning a bit more how to <o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>> administer a Gluster server.), buy you may find it useful.<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>> <o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>> <a href="http://download.gluster.org/pub/gluster/glusterfs/doc/HA%20and%20Load%25"><span style='color:windowtext;text-decoration:none'>http://download.gluster.org/pub/gluster/glusterfs/doc/HA%20and%20Load%</span></a><o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>> 20Balancing%20for%20NFS%20and%20SMB.html<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>> <o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>> Pablo.<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>The thing with this way of failover is though, that you will likely corrupt a currently written file. If your NFS-server (gluster) node dies while you write your file will be corrupt. If you use native glusterfs mounts it will not (should not). This is why I consider the NFS server feature nothing more than a bad hack. It does not deliver the safety that glusterfs promises, even if you solve the failover problem somehow.<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><o:p> </o:p></span></p><p class=MsoPlainText><span lang=EN-US>--<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Regards,<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Stephan<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>_______________________________________________<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US>Gluster-users mailing list<o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><a href="mailto:Gluster-users@gluster.org"><span style='color:windowtext;text-decoration:none'>Gluster-users@gluster.org</span></a><o:p></o:p></span></p><p class=MsoPlainText><span lang=EN-US><a href="http://supercolony.gluster.org/mailman/listinfo/gluster-users"><span style='color:windowtext;text-decoration:none'>http://supercolony.gluster.org/mailman/listinfo/gluster-users</span></a><o:p></o:p></span></p></div></body></html>