sureguy

Members
  • Posts

    914
  • Joined

  • Last visited

  • Days Won

    1

Everything posted by sureguy

  1. I have this strange issue with both the grafana and influxDB docker templates where Community Applications is convinced they're not installed but they are. I've tried recreating them (deleting the xml files and docker container image information) and then redownloading them via community apps and reconfiguring them. I keep my appdata in /mnt/cache/Docker/xxxxxxx which I know isn't standard, but this doesn't break any other docker container configurations. Any thoughts?
  2. Is that helpful from a usability standpoint? To me it seems confusing and unintuitive, perhaps I'm in the minority on that. Aside from that the upgrade was successful
  3. That sounds like data may not be being written to the flash drive. Are you getting errors in the syslog relating to the flash drive?
  4. See here for a possible solution: What happens if you boot using safe mode?
  5. Were I you, if I had an available slot I'd get an Intel nic that has (known) good Linux support. Intel has an important reputation that they need to maintain, more-so than Realtek. I don't know a single person that looks specifically for Realtek nics when building something mission critical. Probably not what you want to hear, I know, but I've been using an Intel nic in Unraid since about 2012 (82574L) without any real issues (jumbo frames seem not to work in Unraid). A card with the same chipset can be had for less than $20 USD via ebay.
  6. Upgraded from RC3 to RC4, without issue.
  7. Still appears to be down for me, same error message as above. Legacy mode appears to work. I'm running 6.6.0-rc3, not sure if that is the cause.
  8. I'm not seeing any of this in syslog or the nginx error.log. Should I be looking for these errors elsewhere?
  9. I was never able to get this sorted out. I was able to boot into a live Ubuntu install and use Jumbo frames properly. Both my onboard NIC (Broadcom BCM57781) and an Intel Corporation 82574L Gigabit Network Connection did not work properly I've subsequently replaced my router with an ER-X which doesn't support Jumbo frames, so no reason to pursue it on my end.
  10. I requested it on their git page - but aside from one other person asking for the same thing there's been no other traction. You could just make a script that you can run after it's updated - you'd have to run it manually, but it's easier than having to remember the commands (apt vs apk etc).
  11. Perhaps have a second post in the locked thread that asks people to "Like" the second post it if they're not experiencing any issues. That way it's possible to track the number of forum posters who are not experiencing issues garnering feed back without cluttering the forums. I'm sure there are many people that wait until a couple people report in that the latest testing release is stable before they install for themselves.
  12. Just a heads up gfjardim (the plugin creator) appears to have updated the preclear plugin for 6.5.0 compatibility. It's available at: https://raw.githubusercontent.com/gfjardim/unRAID-plugins/master/plugins/preclear.disk.plg
  13. To be honest, I wouldn't get too hung up on the terminology - it's been discussed and the naming scheme is unlikely to change. I like how Squid explained it here :
  14. Just a small note that if there is a docker update you will have to perform the above steps again. At least I did after the last docker update.
  15. I successfully precleared a 1TB drive with 3 cycles (pre and post read enabled). Notifications via notification agent worked.
  16. Done: I downloaded both md5 summed them and ran diff against them - here's the output: bishop@ubuntu-pihole:~$ md5sum preclear.new preclear.old 9b38439b09c9a763d50414a4a328485d preclear.new 739681c3298b9be7bacbf1ca5f5fc802 preclear.old bishop@ubuntu-pihole:~$ diff preclear.new preclear.old 395c395 < '7.2.0'|'7.3.0') --- > '7.2.0')
  17. If you look at my post above, it's in there. https://raw.githubusercontent.com/dohlin/unRAID-plugins/master/plugins/preclear.disk.plg
  18. If you install the linked plugin it maintains backward compatibility as it replaces the gcc version line like so: case $gcc_version in '7.2.0'|'7.3.0')
  19. Doesn't seem to. Tmux is probably being used in order to show the progress window (that's the only reason I can imagine that it's needed). And I'm running a preclear right now - and it's being used and outputting progress: root@Phatstore:~# uname -a Linux Phatstore 4.14.16-unRAID #1 SMP PREEMPT Wed Jan 31 08:51:39 PST 2018 x86_64 Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz GenuineIntel GNU/Linux root@Phatstore:~# ps axu | grep tmux root 17733 0.0 0.0 27864 6304 ? Ss 11:28 0:06 /usr/bin/tmux new-session -d -x 140 -y 200 -s preclear_disk_5VP8L3B3 root 27355 0.0 0.0 9764 2108 pts/3 S+ 13:33 0:00 grep tmux Included is a screenshot showing tmux installed by preclear_disk in nerd tools, and the current progress report.
  20. The script appears to use the mail command, which is not installed by default on unRAID (and isn't available from NerdPack). Without that installed email notifications will not work and the command will probably fail. The options in the script would be: -m email@somedomain.com = optional recipient address. If blank and -M option is used, it will default to default e-mail address of "root" -M 1 = Will send an e-mail message at the end of the final results (default if -m is used, but no other -M option given) -M 2 = Will send an e-mail same as 1 plus at the end of a cycle (if multiple cycles are specified) -M 3 = Will send an e-mail same as 2 plus at the start and end of the pre-read, zeroing, post-read -M 4 = Will send an e-mail same as 3 plus also at intervals of 25% during the long tests The -m, -M options requires that a valid working mail command is installed.
  21. Dontlookoverthere over on the unRAID subreddit figured out how to get the plugin running again with a simple edit to the plg: https://www.reddit.com/r/unRAID/comments/7wjpus/preclear_working_on_641/ Caveat emptor and all that. I'm testing it on a drive in my server presently, it seems to be working properly. ohlin5 in the same post has provided an installable version that works with 6.4.1 and should maintain backwards compatiblity here: https://raw.githubusercontent.com/dohlin/unRAID-plugins/master/plugins/preclear.disk.plg
  22. You can't simply take your use case scenario, and then apply it to everyone else. Some people use unRAID simply as a media server and don't necessarily have a PC that can be on for extended periods of time for this work (a road warrior with only a laptop for work for instance). Some people may only have a work laptop, and consume media with hardware other than a PC. Most people won't have an e-sata dock like you have - e-sata was never a very popular technology. Your proposed solution requires a financial investment on the part of some of the (probably less technically savvy) members of the community for something that once required none. If the preclear script can be made to work in unRAID 6.4.1 I really don't see the harm.
  23. To point #2 this assumes the user has a second system that is compatible with the software, and they want to run that system for an extended period of time (presumably without restarting it, etc). There is some advantage to doing a preclear on an unRAID system, especially if the intention is to add the drive to the array if it passes testing.
  24. If you run preclear_disk.sh -l The following error is generated as part of the output: ./preclear_disk.sh: line 236: strings: command not found In regards to the /root/mdcmd issue, I'm not sure when that's called, I only noticed it because the same line also calls strings. Apparently the script still works with these issues - it will preclear the disk. The first instance of the strings command is reading super.dat. Off the top of my head that could be used to discern what drives shouldn't be precleared as they're part of the array.
  25. Looks like unRAID 6.4.1 (not sure about 6.4.0) no longer has strings as a command, but does have strings-BSD which appears to output the same thing. There is also a call to /root/mdcmd which now resides in /usr/local/sbin/mdcmd - The following command will resolve both these issues: sed -i -e "s|strings|strings-BSD|" -e "s|\/root\/mdcmd|\/usr\/local\/sbin\/mdcmd|" preclear_disk.sh