ksarnelli
-
Posts
78 -
Joined
-
Last visited
Content Type
Profiles
Forums
Downloads
Store
Gallery
Bug Reports
Documentation
Landing
Posts posted by ksarnelli
-
-
19 hours ago, ZappyZap said:
I am using https://github.com/loganmarchione/docker-postfixrelay
i could probably do that , but it would require me to build the conatiner myself.
I always priviledge the official container when it existyou could ask on the project see if he is willing to do so.
Thanks, will do.
-
Thanks for packaging this up - I currently have my postfix relay running on a raspberry pi but now I'm setting up this container as a backup. One small ask - can you make the 'mynetworks' setting (in /etc/postfix/main.cnf) configurable as an environment variable? Right now it defaults to 0.0.0.0/0 but I'd like to lock it down to particular hosts.
-
Anyone ever get to the bottom of this? I was hoping some update in 6.10 would fix the issue, but it still persists. In my case it's like a 30+ seconds delay when resizing the docker page.
-
Did you ever figure this out? I have the same issue - it happens whenever the window is resized while the docker page is opened. Ctrl-F adds a search bar to the top, which in turn resizes the page, causing the issue - but you can simply resize the browser window normally to force the issue. Mine hangs for around 30 seconds when this happens.
-
Same issue here.
-
1
-
-
I have this issue too - did anyone ever figure this out?
-
Thanks for this, it's really nice. Would you mind adding an environment variable to set the permissions for the newly created files/folders (umask)?
-
2 hours ago, dlandon said:
Just for a test I created a hidden share on a Windows 10 computer and successfully mounted it with UD.
The share did not have a '$' in the share name. Why are you using a '$'? The C drive will show with a '$' sign, but you shouldn't mount the full C drive anyway for security reasons.How did you create a hidden share on Windows without a $ sign? I'm not doing any drive level sharing, just regular user shares.
-
1 hour ago, dlandon said:
From the UD change log:
2016.04.17 - Fix: Block mounting smb shares with a '$' in the share path name.It's been that way for a long time. No one has made an issue of it until now. It looks to be an edge case to me.
Try to find a work around, like securing the share and not hiding it.
Just for my own sanity, I made the changes I described above, recreated the package archive and deployed it. Everything works fine. I'm able to add $ shares, access them, delete them. The online status remained connected and I was able to delete the $ share without affecting any of my other SMB mounts. If you want me to send lib.php I will happily send it over (and obviously UnassignedDevices.php needed to be modified to remove the $ check as well).
Otherwise I guess my course of action will be to just script these changes every time a new version of unassigned devices is released, but that isn't ideal.
EDIT - I just realized this plugin is on github...I'll just submit a PR.😂
EDIT 2 - PR submitted 😀
-
16 minutes ago, dlandon said:
From the UD change log:
2016.04.17 - Fix: Block mounting smb shares with a '$' in the share path name.It's been that way for a long time. No one has made an issue of it until now. It looks to be an edge case to me.
Try to find a work around, like securing the share and not hiding it.
Who said the shares aren't secure? It's about keeping the share list tidy. No need to show users shares that aren't for users.
-
15 minutes ago, dlandon said:
I can appreciate your enthusiasm, but it is not as simple as that. It is not as much a CIFS issue as a php issue. The '$' is a reserved character in php. I did some testing with your proposed fixes and there were many issues. I remember now why I implemented the block of the '$' character.
In my testing, the '$' character in the mount point caused the ping of the server for online status to fail, and when I removed the remote share with a '$' in the mount point, all the remote shares I had defined were deleted.
The '$' in the mount point is a edge case and not something I care to resolve.
It's obviously your decision since you put in all of the work for the plugin, but I don't really think it's an edge case. Using a trailing $ is the only way to hide shares on a Windows server, and I've seen hundreds (if not thousands) of uses in the wild. Usually to reduce the number of visible shares when a server has shares intended for users and others intended for automated processes.
When I tested the code change above I was only testing that one particular function (get_samba_mounts), but after looking at more code I'm pretty sure the other issues you encountered were from other uses of the parse_ini_file() function without the INI_SCANNER_RAW parameter. The default INI_SCANNER_NORMAL tries to parse values and I don't think you want that. From my tests, parse_ini_file() will return an empty array if a section ends in $ unless you include the INI_SCANNER_RAW - which would probably explain the issues you mentioned.
In any case, whatever you decide, thanks for building this plugin. At some point, if you have a little bit of time can you see if using INI_SCANNER_RAW in every parse_ini_file() call resolves the issues?
-
How do you mount an SMB share that contains a $ character (a hidden share running on Windows)? When I try to add the share using unassigned devices it fails with the following error in the log:
unassigned.devices: Share '//blah/blah$' contains a '$' character.#011It cannot be added.I tried manually adding it to samba_mount.cfg, but it still doesn't work there either.
Adding some keywords because this issue is hard to search for: dollar sign
EDIT: I found an old chain about this issue from 2016 where @dlandon incorrectly believed that 'mount -t cifs' does not support hidden windows shares (shares ending with $). I can fill in some missing details - the issue you were experiencing is that you were trying to connect to the default admin shares on Windows (C$, D$, admin$, etc) - in order to mount those you either need to disable UAC on the Windows side or change a registry value (add DWORD LocalAccountTokenFilterPolicy=1 to HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\System). My point is that hidden shares are certainly supported by the cifs mount command, and in my case I'm not trying to mount one of these 'admin' shares - I'm just trying to mount a regular hidden share which are not affected by that policy. It would be great if you can add support for this.
EDIT2: @dlandon - I did some testing with the unassigned devices plugin code and I think the fix is as simple as removing the '$' check and changing line 1195 in include/lib.php from
$samba_mounts = @parse_ini_file($config_file, true);to
$samba_mounts = @parse_ini_file($config_file, true, INI_SCANNER_RAW);It would be great if you can test/implement this minor change. Thanks!
-
8 hours ago, LakersFan said:
Same here. How did you revert back to 4.2.1-1-05?
Bring up the edit page for the container and change the repository from "binhex/arch-qbittorrentvpn" to "binhex/arch-qbittorrentvpn:4.2.1-1-05". By default the "latest" tag is pulled unless you specify a tag in the repository field.
-
1
-
-
3 hours ago, guythnick said:
It looks like the Privoxy component of this is not working with the latest 4.2.2. update. I have many containers that are now failing searches (sonarr, radarr, jackett) because I am using the http proxy through this container.
The only real clue I can see is in the logs below, but I am not sure if this is related or not:
2020-03-29 17:43:40,645 DEBG 'start-script' stdout output: Sun Mar 29 17:43:40 2020 ERROR: Linux route add command failed: external program exited with error status: 2
Same issue here. I reverted to the 4.2.1-1-05 tag for now.
-
On 12/20/2019 at 10:02 AM, coolspot said:
SuperMicro with LACP LAG stopped working with 6.8.0 upgrade in non-GUI mode. In GUI mode I'm able to get an IP address.
Very odd.
6.7.2 and older had no networking issues with LACP. I'm using a SuperMicro X10-SLF7 which has IMPI, but I believe it's on a dedicated port.
This is exactly the issue I was seeing - I was able to get it working in GUI mode by disabling/re-enabling bonding *and* bridging, but upon rebooting into normal (console) mode it wouldn't work again. Still have not had a chance to investigate further, but just wanted to make sure I noted it here. The only difference is that I'm using an older Supermicro board (an X8DTN). In the meantime I disabled LAG on my switch and in unraid.
-
Updated the container this morning and now getting:
nginx: [emerg] dlopen() "/var/lib/nginx/modules/ngx_stream_geoip2_module.so" failed (Error loading shared library /var/lib/nginx/modules/ngx_stream_geoip2_module.so: No such file or directory) in /etc/nginx/modules/stream_geoip2.conf:1
Checked and the module is in fact missing. Any ideas?
-
16 minutes ago, bonienl said:
I have Unifi running on a custom network (br1) and set a fixed IP address, never experienced disconnected or adopting devices after a Unraid upgrade.
It is working very stable for me.
I guess that's what I'm going to have to do too. I don't like running containers in host mode unless they have dedicated IP. It's just weird because I had zero issues for years with the old container, and immediately had issues when I switched to the new one.
-
On 2/26/2019 at 11:20 AM, Hoopster said:
I don't know if the issue is with the container, but, it is the only thing that has changed recently with my UniFi network.
Every couple of days ALL of my devices go into the disconnected/adopting loop. It is guaranteed to happen if I reboot the server for an unRAID update (currently on 6.7.0 rc5); at least it has both times I have updated OS since moving to this container. I either have to manually run set-inform or change the container to host (maybe I will leave it there) to get the devices to be readopted.
I am still running the same version of UniFI (5.9.29) and the same firmware (the latest on all devices) as I was with the prior container with which I never had these issues through many updates. Again, I don't necessarily know why it would be the container as I do not see other reports of repeating adoption issues, but, it is the only change on my UniFi network, so, I am starting there.
If the answer is "it is NOT the container and something else is broken", can you point me at what that might be?
Latest server log attached
Here's the docker run command:
Command: root@localhost:# /usr/local/emhttp/plugins/dynamix.docker.manager/scripts/docker run -d --name='unifi-controller' --net='host' --log-opt max-size='100m' --log-opt max-file='2' -e TZ="America/Denver" -e HOST_OS="Unraid" -e 'UDP_PORT_3478'='3478' -e 'TCP_PORT_8080'='8080' -e 'TCP_PORT_8443'='8443' -e 'TCP_PORT_8880'='8880' -e 'TCP_PORT_8843'='8843' -e 'UDP_PORT_10001'='10001' -e 'UDP_PORT_1900'='1900' -e 'PUID'='99' -e 'PGID'='100' -v '/mnt/user/appdata/unifi/':'/config':'rw,slave' 'linuxserver/unifi-controller:5.9'I have the exact same issue after switching to the new container (using the 5.9 branch). Did you ever find a solution?
-
Looks like this plugin breaks when an unassigned drive is passed through to a VM. Getting lots of PHP errors after saving.
Hopefully you can get it to work. Really like this plugin over the alternatives. Thanks!
-
4 hours ago, ninthwalker said:
Are you using v2? (with the web interface).
It's set for auto tls, so if the server responds that it doesn't support that then it shouldn't use it.
What are your smtp settings?
port?
auth type? (ie login,plain)Yep, using v2 with the web interface.
I don't see a place to specify auth type. I get the following in the logs when trying to send a test email:
/usr/lib/ruby/2.3.0/openssl/ssl.rb:318:in `post_connection_check': hostname "**removed**" does not match the server certificate (OpenSSL::SSL::SSLError) from /usr/lib/ruby/2.3.0/net/smtp.rb:601:in `tlsconnect' from /usr/lib/ruby/2.3.0/net/smtp.rb:564:in `do_start' from /usr/lib/ruby/2.3.0/net/smtp.rb:521:in `start' from /usr/lib/ruby/gems/2.3.0/gems/mail-2.6.3/lib/mail/network/delivery_methods/smtp.rb:112:in `deliver!' from /usr/lib/ruby/gems/2.3.0/gems/mail-2.6.3/lib/mail/message.rb:252:in `deliver!' from /var/lib/nowshowing/mailAnnouncement.rb:100:in `sendMail' from /usr/local/sbin/announcementreport:87:in `main' from /usr/local/sbin/announcementreport:91:in `<main>'
-
I run an internal Postfix SMTP relay server and it seems as though NowShowingv2 currently requires TLS. Can you add an option to select the transport encryption (in this case, none)? Thanks!
-
12 minutes ago, remotevisitor said:
In Linux the permissions to delete a file depends on the write permission of the folder that contains the file and execute permissions on all parent folders not the file permissions/owner.
Thanks! I thought the file also need to have the write bit set. I'm an idiot.
-
EDIT: Ignore me. Thank you, please drive through.
-------
I have a question/concern about using Recycle Bin and whether or not it has an impact on delete permissions.
Here is my scenario:
- I have a file called test.txt with 600 (rw-------) root:root permissions
- The account I use for SMB (not root) can not read or write test.txt (expected), but it can somehow delete it via SMB!
I don't know exactly how the Recycle Bin plugin works on the backend, but the only thing I can think is that it runs as root and is intercepting the SMB delete command and not enforcing permissions.
Please let me know if this is accurate and/or expected behavior, or if I just have something stranger going on. Thanks!
-
On 2/5/2018 at 5:16 AM, eschultz said:
I was planning to do just that actually.
I think FQDN would work best as long as your TLD is configured properly (most people will be .local anyways, which is the default). Reason is MacOS will resolve better using FQDN over just HOSTNAME. Windows seems to work well with either.
Just checking in to see if this was implemented in 6.5 or if it's planned for a future version? I didn't see it in the release notes. Thanks!
[Support] devzwf - Postfix-Relay
in Docker Containers
Posted
Thanks @ZappyZap - everything checks out!