what is wasting that much RAM if vfs_fruit is enabled? had 4 OOM-killers today after not face a single one in the last 7 years and on a different machine i swa a smbd process eating up 6500 MB on a HP Microserver with 8 GB RAM due copy a 300 GB folder with a Apple client and even not release the memory after the operation was finished root 24301 1.9 22.3 1323692 903600 ? SN 12:29 0:13 /usr/sbin/smbd -D samba-4.3.6-0.fc23.x86_64 -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 181 bytes Desc: OpenPGP digital signature URL: <http://lists.samba.org/pipermail/samba/attachments/20160405/19f10076/signature.sig>
On Tue, Apr 05, 2016 at 12:43:37PM +0200, Reindl Harald wrote:> what is wasting that much RAM if vfs_fruit is enabled? > > had 4 OOM-killers today after not face a single one in the last 7 years and > on a different machine i swa a smbd process eating up 6500 MB on a HP > Microserver with 8 GB RAM due copy a 300 GB folder with a Apple client and > even not release the memory after the operation was finished > > root 24301 1.9 22.3 1323692 903600 ? SN 12:29 0:13 > /usr/sbin/smbd -D > > samba-4.3.6-0.fc23.x86_64just got a private mail from someone else who ran into the same issue. Can you run smbd under valgrind with memcheck? Cheerio! -slow
Am 05.04.2016 um 13:03 schrieb Ralph Boehme:> On Tue, Apr 05, 2016 at 12:43:37PM +0200, Reindl Harald wrote: >> what is wasting that much RAM if vfs_fruit is enabled? >> >> had 4 OOM-killers today after not face a single one in the last 7 years and >> on a different machine i swa a smbd process eating up 6500 MB on a HP >> Microserver with 8 GB RAM due copy a 300 GB folder with a Apple client and >> even not release the memory after the operation was finished >> >> root 24301 1.9 22.3 1323692 903600 ? SN 12:29 0:13 >> /usr/sbin/smbd -D >> >> samba-4.3.6-0.fc23.x86_64 > > just got a private mail from someone else who ran into the same issue. > > Can you run smbd under valgrind with memcheck?not that easy in production and it seems to be unpredictable to trigger at least the problem must exist longer because the case with 8 GB of physical RAM completly eaten was on CentOS 7 with samba-4.2.3-12.el7_2.x86_64 while todays issue happened on samba-4.3.6-0.fc23.x86_64 *maybe* you can reproduce it by copy a really large folder with many files and subfolders over a Mac client - on the other hand today it was triggered by afpd allocating memory and OOM-killer decided the process with largest memory usage is some smbd not sure how smbd did grow that large because on the main fileserver apple clients only use netatalk for now which triggeres "fruit_fset_nt_acl: SMB_VFS_NEXT_FSET_NT_ACL failed" and like messages from time to time -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 181 bytes Desc: OpenPGP digital signature URL: <http://lists.samba.org/pipermail/samba/attachments/20160405/3577ed5b/signature.sig>