similar to: Lstat & Dovecot

In a somewhat unusual circumstance I have been having trouble when making R from source and with "R CMD build dse" forcing the machine to look for an NFS server which was not available, and then stalling. Our system administrator traced this back to the fact that these go through all the directories back to "/" , doing an lstat on everything. He suggests that that seems odd,

Hi Ben, do you know if the smallfile benchmark also does interleaved getdents() and lstat, which is what I found as being the key difference that creates the performance gap (further down this thread)? Also, wouldn't `--threads 8` change the performance numbers by factor 8 versus the plain `ls` and `rsync` that I did? Would you mind running those commands directly/plainly on your cluster to

I found the reason now, at least for this set of lstat()s I was looking at. bup first does all getdents(), obtaining all file names in the directory, and then stat()s them. Apparently this destroys some of gluster's caching, making stat()s ~100x slower. What caching could this be, and how could I convince gluster to serve these stat()s as fast as if a getdents() had been done just before

Hi all, I am on gluster 3.10.5 with one EC volume 16+4. One of the machines go down previous night and I just fixed it and powered on. When glusterfsd processes started they consume all CPU on the server. strace shows every process goes over in bricks directory and do a lstat & readlink calls. Each brick directory is 8TB, %60 full. I waited for 24 hours for it to finish but it did not. I

----- Original Message ----- > From: "Niklas Hamb?chen" <mail at nh2.me> > To: "Ben Turner" <bturner at redhat.com> > Cc: gluster-users at gluster.org > Sent: Sunday, September 17, 2017 9:49:10 PM > Subject: Re: [Gluster-users] Confusing lstat() performance > > Hi Ben, > > do you know if the smallfile benchmark also does interleaved

I did a quick test on one of my lab clusters with no tuning except for quota being enabled: [root at dell-per730-03 ~]# gluster v info Volume Name: vmstore Type: Replicate Volume ID: 0d2e4c49-334b-47c9-8e72-86a4c040a7bd Status: Started Snapshot Count: 0 Number of Bricks: 1 x (2 + 1) = 3 Transport-type: tcp Bricks: Brick1: 192.168.50.1:/rhgs/brick1/vmstore Brick2:

Hi, I have a gluster 3.10 volume with a dir with ~1 million small files in them, say mounted at /mnt/dir with FUSE, and I'm observing something weird: When I list and stat them all using rsync, then the lstat() calls that rsync does are incredibly fast (23 microseconds per call on average, definitely faster than a network roundtrip between my 3-machine bricks connected via Ethernet). But

Nick, What are you planning on running over the shared connection? Database, eMail, File Shares? How many users? How much data? What is your I/O profile? I've worked with 'enterprise' storage most of my career either as a consumer, adviser or provider - can't comment on AoE other than to suggest you look at what are the business & technical goals, how they solve it and what

I'm converting an environment from mandrake to centos, and always try to use the applications standard on the distribution. This means a switch from courier to cyrus. A bit of reading uncovered a potential problem - the site uses firstname.lastname at domain, and I have come across references that Cyrus does not support this. I couldn't find this limitation in the Cyrus

Hi, I plan on upgrading from 1.1.13 to 1.2.4 using an RPM. I have read the wiki [1] and have a question about the mail_location variable. Currently, in 1.1.13, I use: mail_location = maildir:/var/spool/virtual_mailboxes/%h with virtual mailboxes. However, I am planning to change it to use: home=/var/spool/virtual_mailboxes mail_location = maildir:%h as part of the new configuration. Is

I've setup postfix to delivered my mail to a maildir located at /var/spool/virtual_mailboxes/mydomain.com/myuser My postgresql database called 'mails' has a user table setup as follows: CREATE TABLE users ( userid VARCHAR(128) NOT NULL, password VARCHAR(128), realname VARCHAR(128), uid INTEGER NOT NULL, gid INTEGER NOT NULL, home VARCHAR(128), mail VARCHAR(255),

Hi Niklas, Out of interest have you tried testing performance with performance.stat-prefetch enabled? -- Sam McLeod @s_mcleod https://smcleod.net > On 14 Sep 2017, at 10:42 pm, Niklas Hamb?chen <mail at nh2.me> wrote: > > Hi, > > I have a gluster 3.10 volume with a dir with ~1 million small files in > them, say mounted at /mnt/dir with FUSE, and I'm observing

On 17/09/17 18:03, Niklas Hamb?chen wrote: > So far the only difference between `ls` and `bup index` I could observe > is that `bup index` chdir()s into the directory to index, ls doesn't. > > But when I `cd` into the dir and run `ls` without directory argument, it > is still much faster than bup index for each stat(). Hmm, bup uses the fchdir() syscall to go into the target

I've been trying to get dovecot working with postfix with virtual domain setting. I noticed that if I started dovecot with "sudo /usr/sbin/dovecot", everything would work fine. But if I started with "sudo /etc/init.d/dovecot start", I would get the following permission error. I checked "ps -ef | grep dovecot" and didn't find anything different. Sep 29

On 18/09/17 17:23, Ben Turner wrote: > Do you want tuned or untuned? If tuned I'd like to try one of my tunings for metadata, but I will use yours if you want. (Re-CC'd list) I would be interested in both, if possible: To confirm that it's not only my machines that exhibit this behaviour given my settings, and to see what can be achieved with your tuned settings. Thank you!

Well, in part we will see if you get this and I get the return. I JUST launched my new mail server after a few miscues. Actually the OS is Redsleeve6, which is the arm port of Centos 6, and perhaps some of my problems may come down to porting problems. But first, it may be that despite my testing, I missed something in the configurations. First clamd problems: Nov 19 08:09:54 z9m9z

Hi We've recently done some testing with a 3.12 disperse cluster. The performance of filesystem stat calls was terrible, taking multiple seconds. We dumped client side stats to see what was going on and noticed gluster STAT was the culprit. tpcdump shows a STAT call being sent and replied to quite fast, but still the client hangs for multiple seconds before returning. After downgrading

On Wednesday, 26 August 2020 21:15:36 CEST Richard W.M. Jones wrote: > On Wed, Aug 26, 2020 at 06:57:43PM +0200, Pino Toscano wrote: > > Fedora 33 switched the DB of RPM to SQLite, so no more Packages/Name/etc > > files. Because any missing file exception is ignored when checking for > > --if-newer, the lack of the Package files was not properly noticed, so > > the

On 11/08/2013 09:12 AM, Steve Capper wrote: > diff --git a/usr/klibc/lstat.c b/usr/klibc/lstat.c > new file mode 100644 > index 0000000..0282eec > --- /dev/null > +++ b/usr/klibc/lstat.c > @@ -0,0 +1,17 @@ > +#include <fcntl.h> > +#include <unistd.h> > +#include <sys/types.h> > +#include <sys/stat.h> > +#include <sys/syscall.h> > +

Fedora 33 switched the DB of RPM to SQLite, so no more Packages/Name/etc files. Because any missing file exception is ignored when checking for --if-newer, the lack of the Package files was not properly noticed, so the appliance was always considered out out date. Check for the SQLite-based DB first, falling back to the old format. Signed-off-by: Pino Toscano <ptoscano@redhat.com> ---