[DRBD-user] umount /drbdpart takes >50 seconds

[Harald Dunkel]

Hi folks,

using drbd umounting /data1 takes >50 seconds, even though the file
system (ext4, noatime, default) wasn't accessed for more than 2h.
umount ran with 100% CPU load.

# sync
# time umount /data1

real    0m52.772s
user    0m0.000s
sys     0m52.740s


This appears to be a pretty long time. I am concerned that there
is some data sleeping in a buffer that gets flushed only at umount
time.

Kernel is version 4.18.0-0.bpo.1-amd64 on Debian Stretch. drbdutils
is 8.9.10-2. drbd.conf is attached. The bond2 interface used for
drbd synchronization is based upon 2 * 10 Gbit/sec NICs.


Every insightful comment is highly appreciated.

Regards
Harri
-------------- next part --------------
#
# see http://www.drbd.org/users-guide-8.4/re-drbdconf.html
# https://www.linbit.com/en/drbd-sync-rate-controller-2/
#
# /etc/hosts
# ~~~~~~~~~~~
# br0 ("external" network interface)
# 192.168.96.184	srvl060a.example.com srvl060a
# 192.168.96.185	srvl060b.example.com srvl060b
#
# bond2 ("internal" network interface, used for drbd synchronization)
# 10.0.0.2	srvl060a.internal
# 10.0.0.3	srvl060b.internal
#
common {
	disk {
		# on-io-error	detach;			# continue in diskless mode (default)
		fencing	resource-only;			# use fence-peer handler 
		resync-rate	512M;			# synchronization rate over a dedicated line (4 each disk!)
		c-plan-ahead	0;
		al-extents 1237;			# activity log extents
	}

	net {
		protocol C;
		max-buffers	32k;
		max-epoch-size	20000;
		sndbuf-size	256k;
		rcvbuf-size	512k;
		# allow-two-primaries;
		cram-hmac-alg sha256;
		# generated using "openssl rand -base64 30":
		shared-secret "YTbwx3QGiBf2xDrWN+9VppvsrOTXFKRy67x3FAu0";
		after-sb-0pri disconnect;
		after-sb-1pri disconnect;
		after-sb-2pri disconnect;
		rr-conflict disconnect;
	}

	startup {
		wfc-timeout 300;
		degr-wfc-timeout 30;
		outdated-wfc-timeout 30;
	}

	handlers {
		pri-on-incon-degr "echo o > /proc/sysrq-trigger ; halt -f";
		pri-lost-after-sb "echo o > /proc/sysrq-trigger ; halt -f";
		fence-peer "/usr/lib/drbd/outdate-peer.sh on srvl060a.example.com 192.168.96.185 10.0.0.3 on srvl060b.example.com 192.168.96.184 10.0.0.2";
		local-io-error "echo o > /proc/sysrq-trigger ; halt -f";

		initial-split-brain "/usr/lib/drbd/notify-split-brain.sh";
		split-brain "/usr/lib/drbd/notify-split-brain.sh";
	}
}

resource data1 {
	device		/dev/drbd1;
	disk		/dev/disk/by-partlabel/data1;
	meta-disk	internal;
	on srvl060a.example.com {
		address		10.0.0.2:7788;
	}
	on srvl060b.example.com {
		address		10.0.0.3:7788;
	}
	startup {
		become-primary-on srvl060a.example.com;
	}
}

resource data2 {
	device		/dev/drbd2;
	disk		/dev/disk/by-partlabel/data2;
	meta-disk	internal;
	on srvl060a.example.com {
		address		10.0.0.2:7789;
	}
	on srvl060b.example.com {
		address		10.0.0.3:7789;
	}
	startup {
		become-primary-on srvl060b.example.com;
	}
}