Built on openSUSE testing channel live (early-adopters/developers only)

Announcements
45

Bit of an update for whoever might be interested really.

Over the past couple of months I’ve had x86_64 Rocksor 4 running on my home NAS without any issue. Well apart from a motherboard failure, but I can’t point the finger at Rockstor for that one. It is Rock steady (pun intended), reliable and with good performance. No complaints from me or the users (ie my family) for document storage and supplying audio/video media to my LMS/Jellyfin/Icecast server.

I’ve also continued with a fair bit of successful testing on a Pi4 and some further RAID testing in x86 VMs, seeing how Rockstor handles changes to RAID levels, drive loss etc. I believe the newer kernel and BTRFS on Leap is more stable when performing constant tests like this, and more so than the same testing done on the CentOS base which I’ve done in tandem.

I’m not an AD user, so the existing issues with that (which I believe are close to being resolved anyway) don’t stop me from happily using Rockstor 4.

I think my testing has come to a close now, and as I type I’m running one last set of tests changing RAID level from 10 to 1 and other odds and sods in my VMs. As far as I’m concerned it’s fit and forget for me :slight_smile:

One last thing - I’m not a Rockon user (yet), so cannot comment on that area.

Looking forward to showtime :slight_smile:

Geoff

4 Likes
46

Hello,
New user (old linux hand) - successfully built system with a few 3TB drives I had lying around on an old intel i3-540.
Everything looks good - defined an NFS share and a Samba Share
I keep getting a pop-up error window which totally prevents doing anything in the GUI.
image
Would like to know how to fix this…
Going to datatables.net - recommends fixing code…

47

@dvgeek Thanks for the report and glad you got the new DIY installer built and deployed.

Yes this has popped up from time to time in a variety of places. The exact cause is likely a bit of us ‘doing it wrong’ somewhere but we’ve yet to narrow it down or get a reliable reproducer in order to know if we’ve fixed it.

The work around, once it occurs, is to simply use the refresh button in your browser. There after it appears to go away. Not specific to the new build but just some lingering issue we’ve had for a while. Again it’s awaiting a reliable reproducer before we can create an issue where we can test/prove whatever fix is submitted.

Hope that helps.

1 Like
48

Thanks for the quick response @phillxnet
I tried that - clicking on okay / refreshing the browser / etc. but it keeps popping up as soon as I click okay - sometimes I think there could be multiple pop-ups, one behind the other, because it would take multiple clicks on okay for the box to go away - but it again reappears before I click anything else - so preventing any operation in the GUI.

49

Removed all Samba Shares and disabled Samba.
Using only NFS shares - and it has disappeared - not popping up now ( well for at least the last 30 minutes)…

50

Raid 6 (I know officially not supported, but…) - 17 x 3TB disks
3 NFS shares
Tried to give an advanced export option of no_root_squash, and got an “Invalid exports input…”
Had to manually edit the /etc/exports file to add this option.
Also the search feature for the forums does not seem to work (or I am not able to figure it out) - in either Chrome or Firefox browsers…

1 Like
51

Hi @dvgeek, I’ll try to help you with these ones…

I suppose you are referring to using the “Advanced Edits” option/button. Your post actually made me realize we do not cover this function in our documentation so thanks a lot for your feedback. I have now created an issue in our Github docs repository to remedy this:

Now, to get back to your issue, I believe the error you are seeing results from the following part of the code:

As you can see there, this error results from the presence of less than 2 parts in the statement that was used in the Advanced Edit box. As a result, would you be able to detail what you put in this box? Did you input the full export statement or just the no_root_squash option? I suspect the latter given the error you encountered, but let us know so that we can make sure we’re not facing a bug that needs fixing here.

Yes, the search button is unfortunately “hidden” by some sub-optimal css theming. I’ve found a fix for this and need to find the time to implement it; it’s always very risky to alter anything in our forum given its size and how precious its contents are, so I don’t want to rush anything. Bottom line, you will find the search button in the top right corner, to the left of your avatar, it will reveal itself when your cursor hovers over it… see a screenshot below:
image

Hope this helps, and thanks again for your feedback… let us know how it goes.

2 Likes
52

Thank you for your response. Yes, you are quite correct - I put in just no_root_squash without quotes and with quotes. I also tried with and without brackets. So, if I understand this, as the code is expecting two parts in the statement - does that mean that I would have to enter entire line as in /etc/exports - for ex:
/export/MyWkStnBackUp *(rw,async,insecure,no_root_squash) ?

If this is the case, something that would have made this intuitive would be to repeat the automatic entries in the manual export entries box. Then the user would know where to enter the advanced options. Just a suggestion.

Also, in my younger days (now retired), having supported multiple devops in the systems & storage areas - I feel happy if something I noticed & commented on helped contribute to a better functionality/product.

1 Like
53

Yes, that is my understanding (I actually haven’t had time to try that myself, but that’s what I could gather from searching through the forum posts).

Agreed that it’s not really intuitive at the moment, and good idea. As it, the “Advanced Edit” button is more of a “Create a manual export” so we might want to improve that wording while we’re at it. Would you mind creating a Github issue with your finding and idea? Here’s the link to the repo issue tracker: https://github.com/rockstor/rockstor-core/issues This way we can keep track of it and correctly credit it to you :wink:

1 Like
54

I will test it at the first opportunity - could still be after 24 hours as system will be busy backing up media.

I’m not looking for any credit - thank you - just trying to help, if I can. This is not a major issue and can be fixed with documentation. Do not want to burden dev’s with another issue when they are busy readying the product for stable release. I will create it as a feature request - so that the dev’s who look at this can prioritize accordingly. Thanks again.

55

As requested - new issue is open

3 Likes
56

Couple of new observations / issues

  1. Cannot download logs - tried both ways (Download Logs Archive & Through the Logs Reader) - Error message is “Couldn’t download - No file”
  2. The Rockstor log is full and is filled with this message "
[14/Jan/2021 09:20:52] ERROR [system.osi:196] non-zero code(1) returned by command: ['/usr/sbin/btrfs', 'qgroup', 'show', '/mnt2/Data/PlexBackUp']. output: [''] error: ["ERROR: can't list qgroups: quotas not enabled", '']

Not sure if something needs to be set-up. If these are issues - I can move this to the issue tracker - but just wanted to check before doing that.
Thanks

1 Like
57

Hi again, @dvgeek, and thanks a lot for the feedback :wink:

I tried to replicate it on Rockstor-4.0.5 and both ways worked for me… I’m unfortunately not familiar with this area of Rocsktor so I’ll have to dig a bit to have some idea of what could be happening. Do you see the link created when clicking on “Download Logs Archive”? If so, does it look like https://<rockstor-ip>/logs/requested_logs.tgz?

That is the normal and expected behavior when quotas are disabled. We actually had a discussion about improving this as it can be misleading, indeed. The forum thread is below:

… with the corresponding Github issue for reference:
https://github.com/rockstor/rockstor-core/issues/2236#issue-769611226

Sorry I can’t be of more help with the logs issue at the moment…

1 Like
58

Yup - I can see the link - and the browser tries to download the file - but errors out with the message “Couldn’t download - No file”. I could copy and paste the download link - and then I get a 404 Not found.
Tried with a couple of different browsers too - didn’t work.

Too bad about the logs - any way it can turned off or change level of detail ?

59

@dvgeek Hello there.
Re:

We currently only have a command line script to enable or disable debug level of logging and by default that’s disabled. As per the referenced discussion we are due to move the quotas disabled to a debug level rather than it’s current error rating.

The code for where this is defined is indicated in the ‘error’ message you receive so in this case the following:

ERROR [system.osi:196] …

refers to system/osi.py but that is a general process wrapper for all run_command() operations:

As can be seen there is a “log” conditional. So you could for the time being find the ‘problem’ run_command entry and adjust it’s log parameter. There are 3 possible candidates in

You could just change the:

log=True

to

log=False

and that ‘condition’ will no longer be logged.

Alternatively you could just enable quotas on the indicated “Data” pool.

Hope that helps.

2 Likes
60

Took the easy way out - enabled quotas on the pool.
Thanks.

Still not able to download logs though…

61

@dvgeek Thanks for the updated, yes enabling quotas is definitely the easier way around this. And the cost, performance wise, is reducing all the time so that’s good.

Given this is a persistent problem for you, could you start a new forum thread detailing the reproducer for this issue. @Flox has indicated that it works for him in 4.0.5-0 so maybe there is something about your setup or some config change you’ve made that is triggering this. If in a new thread we can focus on just this and once a reproducer is at hand we can open a new GitHub issue with a means to prove any fix proposed. It is a strange one as we haven’t had any other reports to date in that area so would be good to get to the cause of this. Remember to detail the browser you are experiencing this issue with and of course the version of Rockstor you are using.

Hope that helps.

3 Likes
62

Hello!

Starting to get my feet wet with OpenSUSE. I followed all of the instructions for a tumbleweed install. I was able to start the install of Rockstor, but I got one error:

Retrieving: rockstor-4.0.1-0.x86_64.rpm ..............................................................[done (4.5 MiB/s)]
rockstor-4.0.1-0.x86_64.rpm:
    Header V4 RSA/SHA256 Signature, key ID 5f043187: NOKEY
    V4 RSA/SHA256 Signature, key ID 5f043187: NOKEY

Looking for gpg key ID 5F043187 in cache /var/cache/zypp/pubkeys.
Repository Rockstor-Testing does not define additional 'gpgkey=' URLs.
rockstor-4.0.1-0.x86_64 (Rockstor-Testing): Signature verification failed [4-Signatures public key is not available]
Abort, retry, ignore? [a/r/i] (a):

I did an Ignore, and it seems to finish up just fine. After the install, Rockstor failed to start (the service failed). Looking, it was complaining about a missing libzmq.so. I did a quick install of that:

zypper in libzmq5

After doing that, the service started up just fine. I then rebooted, but after the reboot SSH failed to start. In the log I saw:

Feb 08 22:27:02 rocknas sshd[2083]: /etc/ssh/sshd_config line 131: Subsystem 'sftp' already defined.

I opened the config file, and sure enough sftp was defined twice:

# override default of no subsystems
Subsystem      sftp    /usr/libexec/ssh/sftp-server

# This enables accepting locale enviroment variables LC_* LANG, see sshd_config(5).
AcceptEnv LANG LC_CTYPE LC_NUMERIC LC_TIME LC_COLLATE LC_MONETARY LC_MESSAGES
AcceptEnv LC_PAPER LC_NAME LC_ADDRESS LC_TELEPHONE LC_MEASUREMENT
AcceptEnv LC_IDENTIFICATION LC_ALL

# Example of overriding settings on a per-user basis
#Match User anoncvs
#       X11Forwarding no
#       AllowTcpForwarding no
#       PermitTTY no
#       ForceCommand cvs server
###BEGIN: Rockstor SFTP CONFIG. DO NOT EDIT BELOW THIS LINE###
Subsystem       sftp    internal-sftp
AllowUsers root

So it appears that when Rockstor did isn’t install, it added an SFTP line to the sshd_config file, but didn’t first check if such a line was already there. When I commented out the first instance of sftp, it started up just fine.

I’ll keep poking around, and let you know if I come across anything else. Is this thread the best place to report?

1 Like
63

Hi @kupan787,

Thanks a lot for your excellent report, and nice finding! I believe we can be sure of what’s happening here thanks to that.

What you found is indeed related to how we detect and ensure proper config of sshd and apparently the default config shipped in Tumbleweed has now changed.

You are absolutely correct…

Not quite… we actually do check for it, but we currently check for the default config line shipped in Leap (and Tumbleweed at the time of that particular work). As this has now changed, we fail to properly detect it and inactivate it. You can see more details in the corresponding pull request linked below:

As you can see there, the default config in Leap is…

Subsystem   sftp    /usr/lib/ssh/sftp-server

… but thanks to your findings, we now know that it will likely soon ship as:

Subsystem      sftp    /usr/libexec/ssh/sftp-server

This small change in path is enough for us not to detect it and thus failing to inactivate it. We should thus make this part more robust and account for what is seemingly to come in Leap in future releases. Would you mind opening an issue in our Github repo?

Yep, that’s what I would have done to fix it as well :wink:

Thanks a lot for your findings and sharing them, that’s exactly the kind of things we need to keep improving the project and keeping up with upstream changes!

2 Likes
64

@kupan787 Glad to see you experimenting with the ‘Built on openSUSE’ side here.
Re:

Unfortunately we’ve had to stop the Tumbleweed support as some of the required libraries, namely some Python2 ones, are no longer available. Which puts us between a rock and a hard place. Both me and @Flox were quite exited for for Rockstor to be working on Tumbleweed for a while but their dropping of the required libraries that serve our technical dept position was an show stopper really. Hence no rpm’s available past the one your found there. We just couldn’t build it any more. There is however a light at the end of the tunnel on this one. But it’s along tunnel. We have to complete our migration to newer Django, then newer Python. The again back to update our Django. The following 2 issues cover the first 2 parts of this require endeavour before we can again work on Tumbleweed:

and

Both have received some attention and I’m currently working on the latter. But no promises just yet unfortunately as I’m still at the viability assessment phase.

Hence us concentrating only on the Leap 15* variant now. This is a shame as the Tumbleweed builds helped with seeing what was comming down the line on occasions, but all in good time and in the end I hope to have our dependencies in a healthy enough state to server both the stable and edge releases of openSUSE. But with the missing Python 2 libs and the effort to track both OS variants we had to drop the Tumbleweed.

I see @Flox has now addressed your more specific issues re sftp but you are likely to find many more as we are just not testing on Tumbleweed any more. But as stated I really hope to change that once we have all our dependencies in a shiny new state. Then it should again be an all round positive effort to at least try and track the super fast moving Tumbleweed openSUSE variant. And as @Flox points out in the above reply, it’s a useful endeavour if we could only get our dependencies lined up and new enough.

I’m a little surprised actually that you were able to install this rpm. Is this a prior existing Tumbleweed install, as last time I tried I could only build/install our own rpms on Tumbleweeds that had the required, now removed upstream, dependencies already installed. That was the main reason why I stopped building the rpms in fact. Plus without those same rpms available and maintained upstream we can’t offer a Tumbleweed target for he installer either.

Hope that helps. And also note that the Leap variants get pretty aggressive back-ports on the kernel and btrfs so there’s that. Also I hope to start building Leap 15.3 rpms soon, but again my plate is a little full currently so no promises on that either.

1 Like