[Dxspider-support] Help needed - still unable to save user_asc files

Fri Dec 27 14:10:34 CET 2019

Hi

I am probably barking up the completely the wrong tree here but I have experienced something not to dissimilar in a VM environment and it was to do with the amount of cores being used on the CPU

They needed to be allocated to the tasks being run rather than being taken from a shared pool I am no expert here but it took a lot of clever people to sort it out.      

Just a thought as I say probably nothing to do with this issue.

73 Keith GU6EFB

From: Dxspider-support [mailto:dxspider-support-bounces at tobit.co.uk] On Behalf Of Michael Carper, Ph.D. via Dxspider-support
Sent: 27 December 2019 02:57
To: The DXSpider Support list
Cc: Michael Carper, Ph.D.
Subject: Re: [Dxspider-support] Help needed - still unable to save user_asc files

I'm attaching a copy of the server stats over the past 30 days (not sure if the file will come through this reflector).

This is a virtual machine (VM) in the Google Cloud running CentoOS. On a slow day, we'll have no less than 500 users connected. On a busy day, more than 1,200. I've got about 24 linked nodes. In the stats, you can see the CPU utilization never goes over 25% and the disk I/O is always low. I don't have any stats on memory usage... 

image.png

We did try putting mojo on TWO different VMs. The VMs came to a screeching halt... pretty much as soon as the peer nodes connected and about a dozen users on it.

I suppose... I could see if I can add memory to this and see how that works out.

As for permissions... well, I haven't changed them since the time I turned this up about five years ago. Here's what it shows:

image.png

And the group looks like this:

image.png

The files look like this:

image.png

VERY interested in any assistance I can get.

Mike, VK4/WA9PIE

On Thu, Dec 26, 2019 at 2:21 PM Dirk Koopman via Dxspider-support <dxspider-support at tobit.co.uk> wrote:

More than once, probably about six times.

One can rebuild the user file and restart the node. All is then well for a while. I believe I have seen it work for more than one week, but within another week or two it will revert to an empty user file. It MAY be related to spawned jobs which write to the user file thus corrupting the user file, but not every time and, in any case, I can't see any spawn jobs that might do a write. I am reluctant to put in file spin locks but it may become necessary. It MIGHT be memory usage, but there is no objective evidence for a lack thereof. 

Mike seems to have the only instance of "standard" (i.e. non-mojo) DXSpider that can (at least for a time) support 1000+ users on quite a lowly single processor GCloud instance - and seemingly with room for many more. As you know Ian, us lesser mortals start to see serious CPU usage above about 4-500 users (depending on processor). And worse, it appears that the mojo branch runs like a snail on mogodon on an identical instance in the same data centre. Which is the exact opposite of anyone else's experience. 

Mike is unwilling to run his node on dedicated hardware for perfectly reasonable reasons. But otherwise, I am stumped.

Dirk G1TLH

On 26/12/2019 14:35, Michael Carper, Ph.D. via Dxspider-support wrote:

Yes. Actually Dirk also attempted it. No good. 

I was also wondering if it’s a permissions issue. But I think it’s fine. 

I’ll send you images of that tomorrow. 

Mike

On Thu, Dec 26, 2019 at 7:57 PM Ian Maude via Dxspider-support <dxspider-support at tobit.co.uk> wrote:

Also, have you tried rebuilding the user file? 

73 Ian

On 26 Dec 2019, 09:30 +0000, Michael Carper, Ph.D. via Dxspider-support <dxspider-support at tobit.co.uk>, wrote:

For more than a year now, WA9PIE-2 has been unable to save user_asc files.

<image.png>

This began happening without warning and without making any other changes.

As a result, even familiar users who connect get this:

<image.png>

'Most' of the users of this cluster are Ham Radio Deluxe users... and we send the user's info (answers to these questions) to the cluster each time a connection is made... but this is still a major pain.

ANYONE have any thoughts about this?

Mike, VK4/WA9PIE

_______________________________________________
Dxspider-support mailing list
Dxspider-support at tobit.co.uk
https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support

_______________________________________________
Dxspider-support mailing list
Dxspider-support at tobit.co.uk
https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support

_______________________________________________
Dxspider-support mailing list
Dxspider-support at tobit.co.uk
https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support

_______________________________________________
Dxspider-support mailing list
Dxspider-support at tobit.co.uk
https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.tobit.co.uk/pipermail/dxspider-support/attachments/20191227/9c4a26ad/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.png
Type: image/png
Size: 29089 bytes
Desc: not available
URL: <https://mailman.tobit.co.uk/pipermail/dxspider-support/attachments/20191227/9c4a26ad/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image004.png
Type: image/png
Size: 192482 bytes
Desc: not available
URL: <https://mailman.tobit.co.uk/pipermail/dxspider-support/attachments/20191227/9c4a26ad/attachment-0005.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image005.png
Type: image/png
Size: 3269 bytes
Desc: not available
URL: <https://mailman.tobit.co.uk/pipermail/dxspider-support/attachments/20191227/9c4a26ad/attachment-0006.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image007.png
Type: image/png
Size: 58025 bytes
Desc: not available
URL: <https://mailman.tobit.co.uk/pipermail/dxspider-support/attachments/20191227/9c4a26ad/attachment-0007.png>