Hi, We have a host, Solaris 10 10/08 s10s_u6wos_07b on SPARC. SWAP is on ZFS. We allocated two swap devices of 64G each, for a total of around 130G. Over the time, swap space usage is increasing until 96% of used: # zpool list NAME SIZE USED AVAIL CAP HEALTH ALTROOT swap_space 130G 125G 4.94G 96% ONLINE - # zfs list NAME USED AVAIL REFER MOUNTPOINT swap_space 127G 487M 1K none swap_space/swapvol01 64G 2.03G 62.4G - swap_space/swapvol02 63G 1.02G 62.5G - ZFS parameters for swap space volume: # zfs get all swap_space NAME PROPERTY VALUE SOURCE swap_space used 127G - swap_space available 487M - swap_space referenced 1K - swap_space quota none default swap_space reservation none default swap_space compressratio 1.00x - [...] In swap_space, I have two volumes, swapvol01 and swapvol02: # zfs get all swap_space/swap_space_01 NAME PROPERTY VALUE SOURCE swap_space/swapvol01 type volume - swap_space/swapvol01 creation Thu Mar 11 9:22 2010 - swap_space/swapvol01 used 64G - swap_space/swapvol01 available 2.03G - swap_space/swapvol01 referenced 62.4G - swap_space/swapvol01 compressratio 1.00x - swap_space/swapvol01 reservation none default swap_space/swapvol01 volsize 64G - swap_space/swapvol01 volblocksize 8K - swap_space/swapvol01 checksum off inherited from swap_space swap_space/swapvol01 compression off default swap_space/swapvol01 readonly off default swap_space/swapvol01 shareiscsi off default swap_space/swapvol01 copies 1 default swap_space/swapvol01 refreservation 64G local # zfs get all swap_space/swap_space_02 NAME PROPERTY VALUE SOURCE swap_space/swapvol02 type volume - swap_space/swapvol02 creation Tue Jan 18 9:14 2011 - swap_space/swapvol02 used 63G - swap_space/swapvol02 available 1.02G - swap_space/swapvol02 referenced 62.5G - swap_space/swapvol02 compressratio 1.00x - swap_space/swapvol02 reservation none default swap_space/swapvol02 volsize 63G - swap_space/swapvol02 volblocksize 8K - swap_space/swapvol02 checksum off inherited from swap_space swap_space/swapvol02 compression off default swap_space/swapvol02 readonly off default swap_space/swapvol02 shareiscsi off default swap_space/swapvol02 copies 1 default swap_space/swapvol02 refreservation 63G local However, if we do swap -s, we have more than 54GB of free space (swap file name is changed for confidentiality): # swap -s total: 51745880k bytes allocated + 39077200k reserved = 90823080k used, 56984152k available # swap -l swapfile dev swaplo blocks free /dev/swapvol01 256,1 16 134217712 120021888 /dev/swapvol02 256,2 16 132120560 117956320 I understand that "swap -l" only show space used on those devices and "swap -s" include physical memory AND backing store. So logically, we should see a low usage of both swap devices ... ? But, /tmp, /etc/svc/volatile and /var/run directories aren''t used: # df -h /tmp /var/run /etc/svc/volatile Filesystem size used avail capacity Mounted on swap 54G 8.8M 54G 1% /tmp swap 54G 232K 54G 1% /var/run swap 54G 1.6M 54G 1% /etc/svc/volatile We have 20 zones but no one is taking many space in those directories (cut many lines for better reading): # for i in `zoneadm list |grep -v global`; do zlogin $i df -h /tmp; done Filesystem size used avail capacity Mounted on swap 55G 275M 54G 1% /tmp swap 55G 270M 54G 1% /tmp swap 55G 541M 54G 1% /tmp swap 55G 274M 54G 1% /tmp swap 55G 215M 54G 1% /tmp swap 55G 281M 54G 1% /tmp swap 54G 328K 54G 1% /tmp swap 54G 520K 54G 1% /tmp swap 55G 541M 54G 1% /tmp swap 56G 1.4G 54G 3% /tmp swap 54G 576K 54G 1% /tmp swap 54G 328K 54G 1% /tmp swap 54G 432K 54G 1% /tmp swap 55G 270M 54G 1% /tmp swap 54G 40K 54G 1% /tmp swap 54G 4.5M 54G 1% /tmp swap 54G 8K 54G 1% /tmp swap 54G 0K 54G 0% /tmp swap 54G 40K 54G 1% /tmp swap 54G 64K 54G 1% /tmp Any ideas? Because I''m puzzled ... ! Thanks! -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20110802/8c220cba/attachment-0001.html>
What does echo "::memstat" | mdb -k show? max On Aug 2, 2011, at 4:10 PM, Mister Anonyme wrote:> > Hi, > > We have a host, Solaris 10 10/08 s10s_u6wos_07b on SPARC. > > SWAP is on ZFS. > > We allocated two swap devices of 64G each, for a total of around 130G. > > Over the time, swap space usage is increasing until 96% of used: > > # zpool list > NAME SIZE USED AVAIL CAP HEALTH ALTROOT > swap_space 130G 125G 4.94G 96% ONLINE - > > # zfs list > NAME USED AVAIL REFER MOUNTPOINT > swap_space 127G 487M 1K none > swap_space/swapvol01 64G 2.03G 62.4G - > swap_space/swapvol02 63G 1.02G 62.5G - > > > ZFS parameters for swap space volume: > > # zfs get all swap_space > NAME PROPERTY VALUE SOURCE > swap_space used 127G - > swap_space available 487M - > swap_space referenced 1K - > swap_space quota none default > swap_space reservation none default > swap_space compressratio 1.00x - > [...] > > > In swap_space, I have two volumes, swapvol01 and swapvol02: > > # zfs get all swap_space/swap_space_01 > NAME PROPERTY VALUE SOURCE > swap_space/swapvol01 type volume - > swap_space/swapvol01 creation Thu Mar 11 9:22 2010 - > swap_space/swapvol01 used 64G - > swap_space/swapvol01 available 2.03G - > swap_space/swapvol01 referenced 62.4G - > swap_space/swapvol01 compressratio 1.00x - > swap_space/swapvol01 reservation none default > swap_space/swapvol01 volsize 64G - > swap_space/swapvol01 volblocksize 8K - > swap_space/swapvol01 checksum off inherited from swap_space > swap_space/swapvol01 compression off default > swap_space/swapvol01 readonly off default > swap_space/swapvol01 shareiscsi off default > swap_space/swapvol01 copies 1 default > swap_space/swapvol01 refreservation 64G local > > > > # zfs get all swap_space/swap_space_02 > NAME PROPERTY VALUE SOURCE > swap_space/swapvol02 type volume - > swap_space/swapvol02 creation Tue Jan 18 9:14 2011 - > swap_space/swapvol02 used 63G - > swap_space/swapvol02 available 1.02G - > swap_space/swapvol02 referenced 62.5G - > swap_space/swapvol02 compressratio 1.00x - > swap_space/swapvol02 reservation none default > swap_space/swapvol02 volsize 63G - > swap_space/swapvol02 volblocksize 8K - > swap_space/swapvol02 checksum off inherited from swap_space > swap_space/swapvol02 compression off default > swap_space/swapvol02 readonly off default > swap_space/swapvol02 shareiscsi off default > swap_space/swapvol02 copies 1 default > swap_space/swapvol02 refreservation 63G local > > > However, if we do swap -s, we have more than 54GB of free space (swap file name is changed for confidentiality): > > # swap -s > total: 51745880k bytes allocated + 39077200k reserved = 90823080k used, 56984152k available > > # swap -l > swapfile dev swaplo blocks free > /dev/swapvol01 256,1 16 134217712 120021888 > /dev/swapvol02 256,2 16 132120560 117956320 > > I understand that "swap -l" only show space used on those devices and "swap -s" include physical memory AND backing store. So logically, we should see a low usage of both swap devices ... ? > > But, /tmp, /etc/svc/volatile and /var/run directories aren''t used: > > # df -h /tmp /var/run /etc/svc/volatile > Filesystem size used avail capacity Mounted on > swap 54G 8.8M 54G 1% /tmp > swap 54G 232K 54G 1% /var/run > swap 54G 1.6M 54G 1% /etc/svc/volatile > > We have 20 zones but no one is taking many space in those directories (cut many lines for better reading): > > # for i in `zoneadm list |grep -v global`; do zlogin $i df -h /tmp; done > Filesystem size used avail capacity Mounted on > swap 55G 275M 54G 1% /tmp > swap 55G 270M 54G 1% /tmp > swap 55G 541M 54G 1% /tmp > swap 55G 274M 54G 1% /tmp > swap 55G 215M 54G 1% /tmp > swap 55G 281M 54G 1% /tmp > swap 54G 328K 54G 1% /tmp > swap 54G 520K 54G 1% /tmp > swap 55G 541M 54G 1% /tmp > swap 56G 1.4G 54G 3% /tmp > swap 54G 576K 54G 1% /tmp > swap 54G 328K 54G 1% /tmp > swap 54G 432K 54G 1% /tmp > swap 55G 270M 54G 1% /tmp > swap 54G 40K 54G 1% /tmp > swap 54G 4.5M 54G 1% /tmp > swap 54G 8K 54G 1% /tmp > swap 54G 0K 54G 0% /tmp > swap 54G 40K 54G 1% /tmp > swap 54G 64K 54G 1% /tmp > > Any ideas? Because I''m puzzled ... ! > > Thanks! > > > _______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20110802/1c1ce25f/attachment.html>
# echo "::memstat" | mdb -k Page Summary Pages MB %Tot ------------ ---------------- ---------------- ---- Kernel 3258470 25456 40% Anon 4427859 34592 54% Exec and libs 215008 1679 3% Page cache 89353 698 1% Free (cachelist) 72263 564 1% Free (freelist) 172772 1349 2% Total 8235725 64341 Physical 8214142 64172 Subject: Re: [zfs-discuss] swap weirdness From: max at bruningsystems.com Date: Tue, 2 Aug 2011 17:20:49 +0200 CC: zfs-discuss at opensolaris.org To: benetage at hotmail.com What does echo "::memstat" | mdb -k show? max On Aug 2, 2011, at 4:10 PM, Mister Anonyme wrote:Hi, We have a host, Solaris 10 10/08 s10s_u6wos_07b on SPARC. SWAP is on ZFS. We allocated two swap devices of 64G each, for a total of around 130G. Over the time, swap space usage is increasing until 96% of used: # zpool list NAME SIZE USED AVAIL CAP HEALTH ALTROOT swap_space 130G 125G 4.94G 96% ONLINE - # zfs list NAME USED AVAIL REFER MOUNTPOINT swap_space 127G 487M 1K none swap_space/swapvol01 64G 2.03G 62.4G - swap_space/swapvol02 63G 1.02G 62.5G - ZFS parameters for swap space volume: # zfs get all swap_space NAME PROPERTY VALUE SOURCE swap_space used 127G - swap_space available 487M - swap_space referenced 1K - swap_space quota none default swap_space reservation none default swap_space compressratio 1.00x - [...] In swap_space, I have two volumes, swapvol01 and swapvol02: # zfs get all swap_space/swap_space_01 NAME PROPERTY VALUE SOURCE swap_space/swapvol01 type volume - swap_space/swapvol01 creation Thu Mar 11 9:22 2010 - swap_space/swapvol01 used 64G - swap_space/swapvol01 available 2.03G - swap_space/swapvol01 referenced 62.4G - swap_space/swapvol01 compressratio 1.00x - swap_space/swapvol01 reservation none default swap_space/swapvol01 volsize 64G - swap_space/swapvol01 volblocksize 8K - swap_space/swapvol01 checksum off inherited from swap_space swap_space/swapvol01 compression off default swap_space/swapvol01 readonly off default swap_space/swapvol01 shareiscsi off default swap_space/swapvol01 copies 1 default swap_space/swapvol01 refreservation 64G local # zfs get all swap_space/swap_space_02 NAME PROPERTY VALUE SOURCE swap_space/swapvol02 type volume - swap_space/swapvol02 creation Tue Jan 18 9:14 2011 - swap_space/swapvol02 used 63G - swap_space/swapvol02 available 1.02G - swap_space/swapvol02 referenced 62.5G - swap_space/swapvol02 compressratio 1.00x - swap_space/swapvol02 reservation none default swap_space/swapvol02 volsize 63G - swap_space/swapvol02 volblocksize 8K - swap_space/swapvol02 checksum off inherited from swap_space swap_space/swapvol02 compression off default swap_space/swapvol02 readonly off default swap_space/swapvol02 shareiscsi off default swap_space/swapvol02 copies 1 default swap_space/swapvol02 refreservation 63G local However, if we do swap -s, we have more than 54GB of free space (swap file name is changed for confidentiality): # swap -s total: 51745880k bytes allocated + 39077200k reserved = 90823080k used, 56984152k available # swap -l swapfile dev swaplo blocks free /dev/swapvol01 256,1 16 134217712 120021888 /dev/swapvol02 256,2 16 132120560 117956320 I understand that "swap -l" only show space used on those devices and "swap -s" include physical memory AND backing store. So logically, we should see a low usage of both swap devices ... ? But, /tmp, /etc/svc/volatile and /var/run directories aren''t used: # df -h /tmp /var/run /etc/svc/volatile Filesystem size used avail capacity Mounted on swap 54G 8.8M 54G 1% /tmp swap 54G 232K 54G 1% /var/run swap 54G 1.6M 54G 1% /etc/svc/volatile We have 20 zones but no one is taking many space in those directories (cut many lines for better reading): # for i in `zoneadm list |grep -v global`; do zlogin $i df -h /tmp; done Filesystem size used avail capacity Mounted on swap 55G 275M 54G 1% /tmp swap 55G 270M 54G 1% /tmp swap 55G 541M 54G 1% /tmp swap 55G 274M 54G 1% /tmp swap 55G 215M 54G 1% /tmp swap 55G 281M 54G 1% /tmp swap 54G 328K 54G 1% /tmp swap 54G 520K 54G 1% /tmp swap 55G 541M 54G 1% /tmp swap 56G 1.4G 54G 3% /tmp swap 54G 576K 54G 1% /tmp swap 54G 328K 54G 1% /tmp swap 54G 432K 54G 1% /tmp swap 55G 270M 54G 1% /tmp swap 54G 40K 54G 1% /tmp swap 54G 4.5M 54G 1% /tmp swap 54G 8K 54G 1% /tmp swap 54G 0K 54G 0% /tmp swap 54G 40K 54G 1% /tmp swap 54G 64K 54G 1% /tmp Any ideas? Because I''m puzzled ... ! Thanks! _______________________________________________ zfs-discuss mailing list zfs-discuss at opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20110802/e89ef934/attachment-0001.html>