Created attachment 143273 [details] system log showing a problem with dolphin - the kde file system SUMMARY I use a QNAP NAS wit a samba filesystem, the drives are mounted via fstab - see below. Sometimes, but not always, Dolphin gets stuck when writing or reading NAS files. This problem occurs much more in the last one or two months. Also I often get a 1 to 2 -second delay when I first click on a NAS directory in Dolphin. (This occurs mainly in the last month or so separate issue, not topic of this bug report). I realize that this situation difficult to reproduce for the typical developer working on a local filesystem on a laptop. Let me know what extra info I need to provide / what experiments to do to stabilize the network scenario for KDE. I gave the bug a high priority since it prevents the use of KDE in a professional environment with networked storage. STEPS TO REPRODUCE 1. Try to open or save a file on the NAS, mounted with smb, especially after it has fallen asleep 2. 3. OBSERVED RESULT Dolphin stuck, everything using the file system stuck, such as the upload dialog window for creating this bug report. If I open a konsole window then I can normally access the NAS. After a while the whole of KDE freezes and I I cannot shutdown the PC anymore because after a while nothing reacts anymore in KDE, the keyboard does not work anymore, only the mouse cursor still moves. EXPECTED RESULT 1. If the NAS is sleeping I expect a delay of up to 10 seconds for the NAS drives to spin up and the NAS, and thus KDE, to react again. 2. I expect a message from Dolphin that a network storage is not responding / Dolphin and KDE as a whole should never get stuck due to a network problem. (This has always always the main problem for me with KDE since I started using it on a daily basis some 20 years ago) SOFTWARE/OS VERSIONS Windows: macOS: Linux/KDE Plasma: (available in About System) KDE Plasma Version: KDE Frameworks Version: Qt Version: ADDITIONAL INFORMATION The last line shows how the relevant NAS directory is mounted # <file system> <mount point> <type> <options> <dump> <pass> # / was on /dev/sdc2 during installation UUID=440e0d66-6a8d-4c7d-b0e4-7db611d24385 / ext4 errors=remount-ro 0 1 # /boot/efi was on /dev/sdc1 during installation #UUID=C2C4-CCA8 /boot/efi vfat umask=0077 0 1 /swapfile none swap sw 0 0 # ssd data directory, for downloads, thunderbird email, dropbox UUID=d4f31c28-01ee-4252-bde7-671b459317fe /media/ssd_data ext4 rw 0 2 # QNAP TS219P+ //192.168.1.2/data /media/nas_data cifs user=xxxx,password=xxxx,rw,user,nosuid,uid=1000,gid=100,_netdev,vers=1.0 0 0
I configured my NAS to not fall asleep anymore but that was not the cause problem. The problem still occurs, after I log in again after the computer falls asleep. I noticed that I also cannot find any results in krunner that are related to files on the NAS after this happens. When the problem happens, the Dolphin process has status disk-sleep, which always for me is a sign of serious trouble. Even after killing Dolphin and all kio-slaves that are running ( using kill -9) and starting a new Dolphin the problem persists. There are many threads I find when I google ("plasma freezes"), from those I got the idea to try another kernel version and see if that solves the problem. Will report back if that helps.
A while ago I reconfigured the NAS to not fall asleep. Now sometimes, after a suspend and a wakeup, all Dolphin instances are blocked for 1-2 minutes. All blocked instances were showing directories on the NAS. After 2 minutes the instances work again. Here is the part of the syslog from the moment Networkmanager woke up: Jan 02 16:26:29 Big-PC NetworkManager[932]: <info> [1641137189.0177] manager: NetworkManager state is now CONNECTED_GLOBAL Jan 02 16:26:32 Big-PC kwin_x11[1780]: qt.qpa.xcb: QXcbConnection: XCB error: 3 (BadWindow), sequence: 19807, resource id: 81788978, major code: 18 (ChangeProperty), minor cod> Jan 02 16:26:33 Big-PC ovpn-openvpn[1288]: TCP/UDP: Preserving recently used remote address: [AF_INET]95.211.112.9:443 Jan 02 16:26:33 Big-PC ovpn-openvpn[1288]: UDP link local: (not bound) Jan 02 16:26:33 Big-PC ovpn-openvpn[1288]: UDP link remote: [AF_INET]95.211.112.9:443 Jan 02 16:26:39 Big-PC systemd[1]: NetworkManager-dispatcher.service: Succeeded. Jan 02 16:27:12 Big-PC NetworkManager[932]: <warn> [1641137232.3289] dhcp6 (eno1): request timed out Jan 02 16:27:12 Big-PC NetworkManager[932]: <info> [1641137232.3289] dhcp6 (eno1): state changed unknown -> timeout Jan 02 16:27:12 Big-PC NetworkManager[932]: <info> [1641137232.3290] dhcp6 (eno1): canceled DHCP transaction Jan 02 16:27:12 Big-PC NetworkManager[932]: <info> [1641137232.3290] dhcp6 (eno1): state changed timeout -> done Jan 02 16:27:33 Big-PC ovpn-openvpn[1288]: TLS Error: TLS key negotiation failed to occur within 60 seconds (check your network connectivity) Jan 02 16:27:33 Big-PC ovpn-openvpn[1288]: TLS Error: TLS handshake failed Jan 02 16:27:33 Big-PC ovpn-openvpn[1288]: SIGUSR1[soft,tls-error] received, process restarting Jan 02 16:27:36 Big-PC plasmashell[1820]: kf.sonnet.clients.hspell: HSpellDict::HSpellDict: Init failed Jan 02 16:27:36 Big-PC plasmashell[1820]: Connecting to deprecated signal QDBusConnectionInterface::serviceOwnerChanged(QString,QString,QString) Jan 02 16:27:36 Big-PC dbus-daemon[1660]: [session uid=1000 pid=1660] Activating via systemd: service name='org.kde.runners.baloo' unit='plasma-baloorunner.service' requested > Jan 02 16:27:36 Big-PC systemd[1647]: Starting KRunner provider for baloo file indexer... Jan 02 16:27:36 Big-PC systemd[1647]: plasma-baloorunner.service: Control process exited, code=exited, status=1/FAILURE Jan 02 16:27:36 Big-PC systemd[1647]: plasma-baloorunner.service: Skipped due to 'exec-condition'. Jan 02 16:27:36 Big-PC systemd[1647]: Condition check resulted in KRunner provider for baloo file indexer being skipped. Jan 02 16:27:37 Big-PC systemd[1647]: Started Konsole - Terminal.
I managed to solve this by the command 'balooctl disable.' Wow. I changed this bug report's product to frameworks-baloo. FYIP: The hint was the following info in the systemlog ( journalctl -b): Jan 25 13:58:13 Big-PC systemd[1466]: plasma-baloorunner.service: Control process exited, code=exited, status=1/FAILURE Jan 25 13:58:13 Big-PC systemd[1466]: plasma-baloorunner.service: Skipped due to 'exec-condition'. Jan 25 13:58:13 Big-PC systemd[1466]: Condition check resulted in KRunner provider for baloo file indexer being skipped. After this I got the disk-sleep version of Dolphin (It did not start up anymore) and after a while the whole plasma gets stuck.
I think what I am experiencing may be the same thing. I am browsing a cifs mounted directory with Dolphin. I am currently doing some work where I am copying the paths of pdf files into a document. Occasionally when I right click a pdf with the intention to "Copy Location" the Dolphin window freezes. I have two Dolphin windows open, but only one is frozen. In System Monitor I see the two Dolphin processes, one indicating "disk sleep" under CPU% column. After a while it unfreezes, and if I'm not careful I've made some clicks on the frozen window which result in some abrupt changes (e.g accidentally opening a panel). I just tried the solution of `balooctl disable` and that seems to have solved it. I'm on Arch Linux with the latest kde versions. My cifs mount is to a fast SSD array over a 10Gbit interface. I have noticed this "disk sleep" freeze with the same setup, but browsing a folder of images with the info panel open. I was browsing images, and somehow coinciding with this "disk sleep" stuff after killing the dolphin process I ended up with `ls -la` showing a link count of ZERO for one of the files it got stuck on. Had to recover it from snapshot, not sure if related.
still getting occasional dolphin hangs with "disk sleep" when interacting with media on samba mount, even with `balooctl disable` (confirmed with `balooctl status` after a hang)
I attempted to reproduce the three described symptoms using a Samba share on my network. Summary: I can reproduce the first symptom. First: Put the Samba host to sleep and attempt to browse the location using Dolphin. - Mount the Samba share - Open Dolphin, browse to the Samba share - Use VNC to put the remote system to sleep - Observe: Dolphin window becomes unresponsive. Waited for more than one minute. Clicked again, system prompted me to terminate it. This looks like it's still a bug. - Terminate the Dolphin window and re-open. Attempt to open the Samba share. See window go grey, but "Loading folder" and then "Loading canceled" "Could not enter folder /mnt/mount_name" - this looks like expected / desired behavior. -- Second: Dolphin hangs if displaying a folder on the remote host, after sleep and wake of local host. Unfortunately, I can't re-test this myself since sleeping my system causes kernel panics related to the Nvidia drivers. Are you able to re-test this with an up to date system? -- Third: Browse files on the Samba share and right click to bring up the menu. I cannot reproduce this issue. I brought up the right click menu with a couple of dozen files without issue. This problem may be unrelated, or may be specific to that particular Samba host.