Running REST API on RHEL7

[brevans]

Getting started with slurm-web. The documentation says get in touch if you want to work on this. Here I am! My first question is: how easy would it be to run just the REST API in a RHEL environment with all its other dependencies satisfied?

I can run the front-end on a ubuntu host that won't have slurm, but the rest api(s) on clusters will need to run in on RHEL7 if I ever want it to be able to work at our site.

[Monique1970]

I have the same kind of question: I have a PoC for a SLURM cluster running on CentOS (future: RHEL) and installed the slurm-web front-end server on ubuntu (the only available install package as far as I know). The documentation is not clear, but it looks like I need to install an (API) component of slurm-Web on the slurm master and/or the slurm database server (the same in my PoC), to be able to connect the slurm-web front-end to the slurm cluster back-end.

So how do I do that?

[bviviano]

I have the latest git source checkout running (2.2.2) on my RHEL 7.3 cluster running Slurm 16.05 (Entire package, RESTAPI and Dashboard from Apache 2.6). Contact me directly if you're interested in the installation procedure. It takes a little work to download all the correct javascript packages and versions because the documentation is lacking on that front.

[Jsullivan83]

@bviviano I'm interested in your procedure. Would you mind sharing?

[saurabh-cdac]

@bviviano Can you share the installation procedure? Thanks

[bviviano]

@saurabh-cdac I didn't see there was a way to attach a file before.

RHEL7-Slurm-web.pdf

I have attached my directions for getting Slurm-Web up on RHEL7 from source (git clone) to this issue. These (my) directions are specific to my site and are not meant to be step-by-step installation document for a general site install of Slurm-web under RHEL7, use the PDF as more of a guide. Specific differences to watch out for as compared to a standard install of Slurm-web are:

1) I'm running Slurm and Slurm-web on an active/active HA cluster setup under RHEL 7. I have a shared file system called "/atmos" that is mounted on each of my HA cluster management nodes. As a result, I've installed all of my Javascript, Slurm-web, etc files are under "/atmos", but with a similar structure: /atmos/etc /atmos/usr/share /atmos/var/www/html

Then created symlinks over to /etc, /usr/share, etc.  Make sure to adjust any paths with "/atmos" in my document accordingly for your environment, so "/atmos/etc" in my doc becomes "/etc" on your system, "/atmos/usr/share" becomes "/usr/share", etc.

2) Some of the packages (like pyslurm) are specific to the version of Slurm you're running. I'm running Slurm 16.05.8, so adjust as needed for your version of Slurm (Slurm 17 needs a newer version of PySlurm for example then what my directions would download).

3) On the JavaScript PreReqs. Slurm-web is VERY picky about the version of some JavaScript packages. In some cases, downloading the latest packages won't work with Slurm-web. I've noted in my documentation when and what versions matter.

I did my best to walk through the debian package files in the Slurm-web git checkout to figure out what exactly was being done in the .deb and translated that to a manual step that I put in my docs. In some cases I might be copying more files then are absolutely needed (i.e. copying an entire subdirectory for the dashboard, instead of the exact files in that subdirectory that are needed) or combined files (like making a single configuration file for Apache, that included conf, dashboard and REST) mainly as a time saving step.

Do note, I haven't done anything with LDAP authentication inside Slurm-web. LDAP Auth isn't required at my site for viewing the cluster status data, so I haven't looked at it. I see no reason it wouldn't work with any stock LDAP server (slapd, 389, etc), I just haven't tried it under RHEL, so I can't comment.

Everything else from Slurm-web I know works on RHEL7. Every link on my Slurm-web page, from "Jobs" to "Topology" responds correctly, from multiple browsers (Tested IE, Firefox, Chrome and Safari) and multiple OS's (RHEL 6/7 Linux, OS X and Windows 10), so it is possible to get it working on RHEL.

Anyone who has problems with getting this up on RHEL, feel free to ask specific questions, I'll follow up best i can with an answer on this thread, so everyone can see.

[saurabh-cdac]

Thank you so much @bviviano.

[wccropper]

@bviviano I followed your instructions. I have a webpage with a spinning blue arc. I'm on CentOS7.4/SLURM17.11.2. I'm lost as to where to look.

[helterskelter01]

@bviviano I followed your instructions as well but I am not getting this to work. I only get a webpage with a spinning blue arc. Where should I look?

[bviviano]

I have two servers running Slurm-web in a active/active heatbeat cluster. It's gone through many yum upgrades and is currently RHEL 7.4 (Started at 7.3) latest everything. Slurm-web continues to run fine. There are a couple of things you can look at:

1) What version of Slurm are you using? Remember, things like PySlurm are SPECIFIC to the version of Slurm you have installed. If you get the wrong PySlurm for your version of Slurm you have, then the REST API won't work.

My site is still running Slurm 16.05. I haven't done any testing with any Slurm 17 release with Slurm-web. We're planning an upgrade to Slurm 17 sometime later in the summer on our Cluster, so I can't really comment on Slurm 17 from personal experience. Others have reported to me that Slurm 17 does work with Slurm-web on RHEL/CentOS 7 using my instructions, as long as they have the correct PySlurm modules.

2) Check the REST API. https://edf-hpc.github.io/slurm-web/api.html Verify you can get the raw data from the REST API via a web browser. On either of my hosts running slurm-web, I can goto the REST API URL. For example:

    https://slurm-web.localdomain.com/slurm-restapi/nodes

And get a list of all the nodes inside Slurm. My systems show: { "a0n01": { "alloc_cpus": 0, "alloc_mem": 0, "arch": "x86_64", "boards": 1, "boot_time": 1519999934, "core_spec_cnt": 0, "cores": 16, ....

   https://slurm-web.localdomain.com/slurm-restapi/cluster

{ "authentication": { "enabled": false, "guest": false }, "data": { "cores": 47120, "name": "clustername", "nodes": 1460 } } And get the basic details of the cluster, etc.

If the REST API isn't returning correct (or any) data, then the rotating blue ball is going to be your most likely result on the GUI. Go through the REST API HTML Reference page and test the API functions, make sure you get a result before moving onto the GUI debugging.

3) Review the Apache error logs. Anything that's broken inside Slurm-web that isn't client side JavaScript will show up in the Apache error logs. The messages are not always clear, but they do let you know there is a problem on the server side vs. the client side.

4) My systems are running Slurm-web on the tagged 2.2.2 branch. I can't comment at all on any patches/updates that have been released since then and what impact they might have on RHEL/CentOS 7. I suggest you download and try the 2.2.2 tagged branch instead of using the latest Git Clone:

 https://github.com/edf-hpc/slurm-web/releases

There have been some code changes and checkins according to the history since 2.2.2, possibly something has changed in the main repository since I wrote those instructions that no longer works correctly with RHEL/CentOS 7 with my instructions.

My current environment is:

RHEL 7.4 (latest as of 2 weeks ago yum updates) Slurm 16.05 (And associated Python/JavaScript dependencies) Slurm-web 2.2.2 (https://github.com/edf-hpc/slurm-web/archive/v2.2.2.tar.gz)

Hope this helps.

[s0nghuiming]

Do you still have a plan to provide a RH/CentOS supported release? If no, I have to implement my own web project even if it is tiny and weak on many aspects.

[bviviano]

I am still running Slurm-web 2.2.2 on my now RHEL 7.6 cluster. So, the procedure I documented back into 2017 still works. I haven't tried to do anything with 2.2.3, but that's just because I haven't had time. Maybe later this year I'll try an upgrade to 2.2.3 (or 4 if it's out).

[helterskelter01]

I tried your method @bviviano and it worked for me as well. Thank you for your documentation. I used slurm 17.02.7, RHEL 7.6 and slurm-web 2.2.2.

[diogocalgaro]

3. On the JavaScript PreReqs. Slurm-web is VERY picky about the version of some JavaScript packages. In some cases, downloading the latest packages won't work with Slurm-web. I've noted in my documentation when and what versions matter.

Can you tell what versions of these javascript packages you used? It has been several months from your pdf instruction.

I'm using:

python-pyslurm-17.11-1 bootstrap-4.3.1 bootstrap-tagsinput-master Bootstrap-3-Typeahead-master async-master d3-v3.5.17 requirejs-2.3.6 jquery-2.2.4 tablesorter-2.17.8 flot-0.8.3 handlebars-4.1.1 threejs-master popper.js-1.14.7

Thanks for the pdf!

[marseaplage]

Hello I got these errors:

[Thu Apr 11 15:29:54.223805 2019] [authz_core:error] [pid 11484] [client ip] AH01630: client denied by server configuration: /var/www/html/nodes [Thu Apr 11 15:29:54.514279 2019] [:error] [pid 11482] WARNING: ConnectionError from Redis, server unreachable /usr/sbin/httpd: symbol lookup error: /usr/lib64/slurm/auth_munge.so: undefined symbol: slurm_debug [Thu Apr 11 15:29:54.525787 2019] [core:error] [pid 11484] [client ip] End of script output before headers: slurm-web-restapi.wsgi [Thu Apr 11 15:29:55.978478 2019] [authz_core:error] [pid 11485] [client ip] AH01630: client denied by server configuration: /var/www/html/nodes [Thu Apr 11 15:29:56.232374 2019] [:error] [pid 11507] WARNING: ConnectionError from Redis, server unreachable /usr/sbin/httpd: symbol lookup error: /usr/lib64/slurm/auth_munge.so: undefined symbol: slurm_debug [Thu Apr 11 15:29:56.241709 2019] [core:error] [pid 11485] [client ip] End of script output before headers: slurm-web-restapi.wsgi

Apache 2.4.6 slurm-17.02.4 slurm web 2.2 Centos 7.3

ANy help?

[bviviano]

Without knowing exactly what your Apache configuration file(s) look like, I can only make a guess. But, it appears you have multiple issues

[Thu Apr 11 15:29:54.223805 2019] [authz_core:error] [pid 11484] [client ip] AH01630: client denied by server configuration: /var/www/html/nodes

Indicates that your Apache configuration file isn't allowing access from a browser.

https://linux.101hacks.com/unix/client-denied-by-server-configuration/

When you browse to the Slurm-web install point, do you see anything, even the basic static images?

[Thu Apr 11 15:29:54.514279 2019] [:error] [pid 11482] WARNING: ConnectionError from Redis, server unreachable /usr/sbin/httpd: symbol lookup error: /usr/lib64/slurm/auth_munge.so: undefined symbol: slurm_debug

Do you have Munge and all the other pieces needed to talk to Slurm installed on the system you're running Slurm-web from? The system hosting Slurm-web needs to be able to talk with your slurmctld daemon just like any other node of the cluster.

The easiest way to check is install the Slurm utilities like sinfo/squeue/scontrol onto your Apache box and verify they work.

[bviviano]

Also, The message:

[Thu Apr 11 15:29:56.232374 2019] [:error] [pid 11507] WARNING: ConnectionError from Redis, server unreachable

Can be ignored.

$ grep "ConnectionError from Redis" /var/log/httpd/error_log-20190407 | wc -l 204340

I have thousands of such messages each week in my Apache error logs. Redis is something Python specific, I don't really know what it's being used for inside Slurm-web and I am not a Python programmer. Whatever it is, it doesn't stop Slurm-web from working on RHEL/CentOS.

[marseaplage]

Hello @bviviano thank you so much for your guidance. Right now I just got this in my apache configuration file:

vi slurm-web.conf

<VirtualHost *:8084> WSGIDaemonProcess slurm-web-restapi user=apache group=apache threads=5 WSGIScriptAlias /slurm-restapi /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi WSGIPassAuthorization On

<Directory /usr/share/slurm-web/restapi/> WSGIProcessGroup slurm-web-restapi WSGIApplicationGroup %{GLOBAL} Options Indexes FollowSymLinks Require all granted

But I still get this error

[Fri Apr 12 10:56:08.154003 2019] [authz_core:error] [pid 44294] [client ip] AH01630: client denied by server configuration: /var/www/html/slurm-restapi

It looks like the wsgi process never run or the Alias is ignored, because actually it should access to /usr/share/slurm-web/restapi/ not to /var/www/html. Any idea??

[bviviano]

When using a VirtualHost Entry, there are other things you may need to define to make the Virtual Host work correctly. I suspect you're missing one or more directives that are defined in the main /etc/httpd/conf/httpd.conf, that are missing from your VirtualHost that is preventing Slurm-web from working. As I have never setup Slurm-web in an Apache VirtualHost and this is not a general discussion group about how to configure Apache VirtualHost's, I can't really offer you any insight there.

When dealing with issues like this, I always recommend you simplify your setup. Get rid of the VirtualHost part of your configuration and set it up on good old port 80. Once you can make it work with the a Apache configuration and port 80, then you can try and make it work inside a VirtualHost on a different port.

[marseaplage]

Could you please tell me which version of every package you have installed on your server? Because I notice that my problem is because the last version of mod_wsgi that I am using is not compatible with the version of python that I am using (2.7) and I can not update because of pyslurm for my slurm is using it.

[bviviano]

I can't give you a list of every package on my system. Did you compile your own mod_wsgi or did you use the one from the yum repo? All my base packages for Apache/Python are straight from either the RHEL or EPEL repositories:

$ rpm -q mod_wsgi httpd python mod_wsgi-3.4-18.el7.x86_64 httpd-2.4.6-88.el7.x86_64 python-2.7.5-76.el7.x86_64

If you're using the RHEL or CentOS Apache/mod_wsgi, Python along with PySlurm 17.02 (since you list you're using Slurm 17.02) you should have no problems with RHEL 7.x.

https://github.com/PySlurm/pyslurm/tree/17.02.0

From README for PySlurm 17.02.0:

This PySlurm branch has been tested with: Cython 0.15.1*, 0.19.2, and the latest stable Python 2.6, 2.7, and 3.4 Slurm 17.02.0 thru 17.02.10

[marseaplage]

Hi @bviviano , I have almost the same version of the packages as you:

$ rpm -q mod_wsgi httpd python mod_wsgi-3.4-12.el7_0.x86_64 httpd-2.4.6-45.el7.centos.4.x86_64 slurm-17.02.4-1.el7.centos.x86_64 python-2.7.5-48.el7.x86_64

y cython 0.25.2

Theoretically must work, but it does not work, now I have this error:

http://server:8084/slurm-restapi

Not Found

The requested URL was not found on the server. If you entered the URL manually please check your spelling and try again.

And my apache configuration in slurm-web.conf is: (just for testing purposes:

WSGIDaemonProcess slurm-web-restapi user=apache group=apache threads=5 WSGIScriptAlias /slurm-restapi /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi WSGIPassAuthorization On

WSGIDontWriteBytecode On

<Directory /usr/share/slurm-web/restapi/>

# Apache 2.4 Require all granted

<IfModule !mod_authz_core.c>

Apache 2.2

Order Deny,Allow
Allow from all

<VirtualHost *:8084> #

/slurm-restapi

#

<Directory /usr/share/slurm-web/restapi/> WSGIProcessGroup slurm-web-restapi WSGIApplicationGroup %{GLOBAL} Options Indexes FollowSymLinks

Any help about that?

[bviviano]

Just wanted to drop by and give an update. I've completed an upgrade of Slurm from 16.05.08 -> 17.02.11 -> 17.11.13-2, on my way up to 19.05.

I can confirm that Slurm-web tagged release 2.2.2 is working fine with RHEL 7.6 and all 3 versions of Slurm under RHEL 7.6.

I am using Slurm-web 2.2.2 because I ran into some issues with 2.2.3 and PySlurm versions on 16.05 and 17.02 related to new Slurm features not supported on the older releases. I plan to try 2.2.3 once I get everything up to 18.08 in a month or so.

As before, the key thing is to make sure your version of PySlurm matches your version of Slurm, or Slurm-web won't be happy.

Also, on RHEL7 systems, starting with PySlurm 17.11, you'll get an error in Apache related to the Python Egg Cache:

[Errno 13] Permission denied: '/usr/share/httpd/.python-eggs'

Two options to fix this:

First, just create /usr/share/httpd/.python-eggs and make it owned by the Apache User:

$ mkdir /usr/share/httpd/.python-eggs $ chown apache.apache /usr/share/httpd/.python-eggs

Second, and more RHEL like, you can "extend" the systemd httpd service to set the environment variable PYTHON_EGG_CACHE to point at someplace the Apache user can write to:

$ /bin/mkdir /etc/systemd/system/httpd.service.d $ /bin/cat > /etc/systemd/system/httpd.service.d/pythonegg.conf << EOF [Service] Environment='PYTHON_EGG_CACHE=/tmp' EOF $ /bin/systemctl daemon-reload $ /bin/systemctl restart httpd

Adjust the value of "PYTHON_EGG_CACHE" to where ever makes sense for your system, it just has to be a directory that the "apache" user can write to.

I continue to monitor this thread, so if any RHEL/CentOS users are having problems with Slurm-web, just ask a question and I'll do my best to help.

[BSCrumpton]

@bviviano, I can confirm working slurm 19.05 with pyslurm branch 19.05.0 and slurm-web of some version from about Nov 20 2017. Thank you so much for your PDF way back when, we used it at work for our cluster and it's been chugging along ever since.

[levidd]

@BSCrumpton could you share what js package versions you were using? I am struggling to get my set up running, and don't know where to begin looking.

[BSCrumpton]

@levidd I can check Tuesday when I get back to work. Do you have a command I could run to get you the information you need?

[bmk-star]

@bviviano i am using RHEL 7.7, slurm 18.08.8, pyslurm 18.08.8, beegfs file system and also implement every step on the pdf. but after restarting the httpd, when i try to access the slurm-web i can only see page is loading. ....can any one help

[bviviano]

@bviviano i am using RHEL 7.7, slurm 18.08.8, pyslurm 18.08.8, beegfs file system and also implement every step on the pdf. but after restarting the httpd, when i try to access the slurm-web i can only see page is loading. ....can any one help

I'm currently at Slurm 19.05.5 and RHEL 7.7 and it's working fine. I've been through every rev of Slurm since 16.05 with this setup, so it's not an issue of OS compatability. It's likely you've gotten something wrong in your httpd.conf setup or an incompatable version of one of the dependencies.

Start with, can you access the REST API?

https://edf-hpc.github.io/slurm-web/api.html

Assuming you WSGIScriptAlias is set to the default /slurm-restapi

What happens when you open a browser to the following:

http://cluster.mydomain.com/slurm-restapi/cluster http://cluster.mydomain.com/slurm-restapi/nodes

If you get a reasonable looking result that shows you expected information about your cluster, then it's probably an issue with one of the dependencies for the web interface.

[bviviano]

Wanted to add a comment to this for others benefit.

I just completed the upgrade from Slurm-Web-2.2.2 to 2.2.5 using the slurm-web-debian-2.2.5.tar.gz located on the releases page of this project on both of my Slurm based clusters.

My current environment on both is:

RHEL 7.7 (3.10.0-1062.18.1.el7.x86_64) httpd 2.4.6-90 Slurm 19.05.5 PySlurm 19.05 Slurm-web-2.2.5

Later this summer we're planning an upgrade to Slurm 20.02.

The instructions I uploaded earlier in the case are still valid. If you run into any problems/issues it's most likely related to a dependency (bootstrap*, jquery, d3, etc).

[bmk-star]

@bviviano i can not access the REST API

[bviviano]

@bviviano i can not access the REST API

Then you need to provide more information, I can't see your browser, I can't see your httpd log files.

Start with something simple. On the host you're running the httpd process on, can PySlurm talk to the Slurm API?

[root@cluster-mgmt1 ~]# cat test.py

!/usr/bin/python

import pyslurm

print "PySLURM\t%s" % (pyslurm.version()) print "SLURM\t%s-%s-%s\n" % (pyslurm.slurm_api_version())

[root@cluster-mgmt1 ~]# ./test.py PySLURM 19.05.0.0 SLURM 19-5-5

If the above doesn't work then you have a bigger problem with Slurm that is outside Slurm-Web and can't be resolved in this thread.

If the above works, then you need to start going through your httpd logs and looking for errors related to PySlurm and WSGI.

If you find specific error messages you don't understand, post the details and someone will try and help you resolve them.

[bmk-star]

@bviviano test.py works and return PySLURM 18.08.8.1 SLURM 18-8-8 And the httpd log has is [core:error] [pid 10992] [client ::1:49264] End of script output before headers: slurm-web-restapi.wsgi The browser also looks like this For this http://cluster.mydomain.com/slurm-restapi/cluster Internal Server Error The server encountered an internal error or misconfiguration and was unable to complete your request. Please contact the server administrator at root@localhost to inform them of the time this error occurred, and the actions you performed just before this error. More information about this error may be available in the server error log.

For this http://cluster.mydomain.com/slurm-restapi Not Found The requested URL was not found on the server. If you entered the URL manually please check your spelling and try again.

[bmk-star]

@bviviano Thanks for the help i really appreciate

[bviviano]

What version of Slurm-web are you using? I am currently using 2.2.2 from https://github.com/edf-hpc/slurm-web/archive/v2.2.2.tar.gz. I found that the 2.2.3 dashboard doesn't work for me and while 2.2.4/2.2.5 display, both have a bug related to the dashboard ([https://github.com/edf-hpc/slurm-web/issues/210]) that causes single core jobs to not render correctly, so I moved back to 2.2.2. The 2.2.2 version with matching PySlurm has worked reliably for me for Slurm 16.08 -> 19.05.

Did you follow my earlier instructions and install everything under /usr/share/slurm-web? There are places in the Slurm-web source code that hard code the /usr/share/slurm-web path? If you installed the Slurm-web source code some place other then /usr/share/slurm-web, it won't work. The only way you can install it into location other then /usr/share/slurm-web is you MUST create a symlink to /usr/share/slurm-web to where you installed it (that's what I did as I have a clustered httpd setup with failover on a shared drive):

[root@cluter-mgmt1 ~]# ls -l /usr/share/slurm-web lrwxrwxrwx. 1 root root 25 Jun 22 2017 /usr/share/slurm-web -> /shared/www/html/slurm-web

[root@cluster-mgmt1 ~]# ls -l /shared/www/html/slurm-web/ total 5 drwxr-xr-x. 2 root root 4096 Jun 17 2019 conf-server drwxr-xr-x. 5 root root 4096 May 9 2019 dashboard drwxr-xr-x. 3 root root 4096 Mar 26 12:00 restapi

If you do that, make sure you include

Options FollowSymLinks

in all your httpd 2.4 directives.

Assuming you did follow my earlier install directions and are still having problems, what is the permission settings on your slurm-web-restapi.wsgi and other files under /usr/share/slurm-web?

[root@cluster-mgmt1 ~]# grep slurm-web-restapi.wsgi /etc/httpd/conf.d/slurm-web.conf WSGIScriptAlias /slurm-restapi /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi

[root@cluster-mgmt1 ~]# ls -l /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi -rw-rw-r--. 1 root root 154 Apr 27 2017 /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi

[root@cluster-mgmt1 ~]# ls -l /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi -rw-rw-r--. 1 root root 154 Apr 27 2017 /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi

[root@cluster-mgmt1 ~]# cat /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi import logging, sys logging.basicConfig(stream=sys.stderr) sys.path.insert(0, '/usr/share/slurm-web/restapi') from slurmrestapi import app as application

Make sure that everything (files and directories) under /usr/share/slurm-web is readable by the httpd/apache user.

[bmk-star]

@bviviano i used slurm-web 2.2.2 also and follow every instruction on the pdf you attached. i also used a shared file system with a mount point /beegfs/. [centos@hpc_master ~]$ ls -l /usr/share/slurm-web lrwxrwxrwx 1 root root 36 Mar 26 17:37 /usr/share/slurm-web -> /beegfs/atmos/www/html/slurm-web

[centos@hpc_master ~]$ ls -l /beegfs/atmos/www/html/slurm-web total 2 drwxr-xr-x 2 root root 10 Mar 27 16:13 conf-server drwxr-xr-x 5 root root 5 Mar 26 17:17 dashboard drwxr-xr-x 3 root root 15 Mar 27 16:20 restapi

[centos@hpc_master ~]$ grep slurm-web-restapi.wsgi /etc/httpd/conf.d/slurm-web.conf WSGIScriptAlias /slurm-restapi /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi

[centos@hpc_master ~]$ ls -l /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi -rw-rw-r-- 1 root root 154 Apr 27 2017 /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi

[centos@hpc_master ~]$ cat /usr/share/slurm-web/restapi/slurm-web-restapi.wsgi import logging, sys logging.basicConfig(stream=sys.stderr) sys.path.insert(0, '/usr/share/slurm-web/restapi') from slurmrestapi import app as application

and still getting the same error [core:error] [pid 31254] [client ip] End of script output before headers: slurm-web-restapi.wsgi any one help

[bmk-star]

@bviviano Thanks for the help, i solve it. It's a permission issue but not on the slurm-web files or directories, it on the installed slurm. its permission was only for root user. i give read and execute permission for the group and other user, it works fine.
Now i'm trying to solve the slurm dashboard issue it only show the basic static images

[bviviano]

Make sure the Apache user has read access to everything and "tail -f" all the apache logs while you try and access the dashboard. The dashboard is the hard part, getting the correct versions of all the dependencies can be tricky and resolving any WSGI problems with Python.

[bmk-star]

@bviviano ls -l /usr/share/slurm-web drwxr-xr-x 2 root root 4096 Apr 3 14:36 conf-server drwxr-xr-x 5 root root 74 Apr 3 14:21 dashboard drwxr-xr-x 3 root root 4096 Apr 6 15:29 restapi

If you see my dashboard file its size is 74, but yours is 4096. may be this is the problem. any idea?

[bviviano]

That's just the directory size.

What do you see in the Apache Access and Error logs when you try and access the web site?

[bmk-star]

it doesn't return any thing... i don't know why

[bviviano]

You're not seeing ANYTHING in the Apache Error and Access logs under /var/log/httpd?, not even the attempt to access the /slurm URL? If so, then there is a problem with your Apache setup.

[bmk-star]

sorry .... it doesn't return anything for the error log but for the access log it returns [06/Apr/2020:17:18:14 +0000] "GET /javascript/bootstrap/css/bootstrap.css HTTP/1.1" 304 - "http://mydomain/slurm/" "Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Firefox/68.0"

[06/Apr/2020:17:18:14 +0000] "GET /slurm/css/dashboard.css HTTP/1.1" 304 - "http://mydomain/slurm/" "Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Firefox/68.0"

[06/Apr/2020:17:18:14 +0000] "GET /slurm-web-conf/clusters.config.js HTTP/1.1" 304 - "http://mydomain/slurm/" "Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Firefox/68.0"

[06/Apr/2020:17:18:14 +0000] "GET /javascript/bootstrap/css/bootstrap-tagsinput.css HTTP/1.1" 304 - "http://mydomain/slurm/" "Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Firefox/68.0"

[06/Apr/2020:17:18:14 +0000] "GET /javascript/requirejs/require.min.js HTTP/1.1" 304 - "http://mydomain/slurm/" "Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Firefox/68.0"

[06/Apr/2020:17:18:14 +0000] "GET /slurm/index.js HTTP/1.1" 404 212 "http://mydomain/slurm/" "Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Firefox/68.0"

[bviviano]

Then the issue is likely a problem with the version of the files you have installed for Javascript. You might have a newer/older version of something then what slurm-web supports.

I am attaching a zip file of the javascript directory from my slurm-web deployment.

I am providing it as an example of what is working for me with Slurm-web-2.2.2 (and 2.2.5).

You should ALWAYS download software from known/trusted sources/sites. This zip file IS NOT a known/trusted package and I make NO claims that it is.

You should scan it before unpacking it and use it ONLY to compare to what you already have setup in your /var/www/html/javascript directory to see what's different, you should NOT simply place my javascript directory into your Apache setup. It is provided so that you (or anyone else having problems) can better understand what's going on with javascript and what might be missing from a setup that isn't working.

javascript.zip

[levidd]

@bviviano Thank you for your help and @bmk-star for discussing. I was able to fix my issues by realizing that slurm-web was written in python2 and I was using python3. Easy mistake that I completely glossed over.

[bmk-star]

@bviviano Thank you so much...

[bviviano]

Posting an update for anyone subscribed to this thread. I just rev'd my testing cluster to:

RHEL 7.9 Slurm 20.02.6 Slurm-web 2.2.6 w/ the patches provides by @BSCrumpton explained in issue #210

Everything is working fine (So far).

[adamya21]

Hi @bviviano , Thankyou so much for providing pdf file for configuration But I'm having trouble installing it in my centos 7 vm

rpm -q mod_wsgi httpd python slurm mod_wsgi-3.4-18.el7.x86_64 httpd-2.4.6-97.el7.centos.x86_64 python-2.7.5-90.el7.x86_64 slurm-19.05.2-1.el7.x86_64

Here the javacript package version that I've installed Are there any mistakes in these? because I'm getting a blue loading symbol in web server.

bootstrap = bootstrap-3.3.7
bootstrap-tagsinput bootstrap-typeahead async d3 = 3.5.7 requirejs = 2.3.6/, text= latest jquery = 2.2.4 jquery-tablesorter = 2.31.3 jquery-flot = 4.2.1 handlebars = v4.7.6 Three= latest

access_log :

192.168.48.1 - - [09/Feb/2021:18:24:46 +0530] "GET /slurm/static/favicon.ico HTTP/1.1" 200 1150 "http://192.168.48.129/slurm/" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.146 Safari/537.36" 192.168.48.1 - - [09/Feb/2021:18:24:46 +0530] "GET /javascript/handlebars/handlebars.js HTTP/1.1" 404 233 "http://192.168.48.129/slurm/" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.146 Safari/537.36" 192.168.48.1 - - [09/Feb/2021:18:24:46 +0530] "GET /slurm/ HTTP/1.1" 304 - "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.146 Safari/537.36" 192.168.48.1 - - [09/Feb/2021:18:24:46 +0530] "GET /slurm/static/favicon.ico HTTP/1.1" 200 1150 "http://192.168.48.129/slurm/" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.146 Safari/537.36" 192.168.48.1 - - [09/Feb/2021:18:24:46 +0530] "GET /javascript/handlebars/handlebars.js HTTP/1.1" 404 233 "http://192.168.48.129/slurm/" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.146 Safari/537.36" ::1 - - [09/Feb/2021:18:24:48 +0530] "OPTIONS HTTP/1.0" 200 - "-" "Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips mod_wsgi/3.4 Python/2.7.5 (internal dummy connection)" ::1 - - [09/Feb/2021:18:24:49 +0530] "OPTIONS HTTP/1.0" 200 - "-" "Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips mod_wsgi/3.4 Python/2.7.5 (internal dummy connection)" ::1 - - [09/Feb/2021:18:24:50 +0530] "OPTIONS * HTTP/1.0" 200 - "-" "Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips mod_wsgi/3.4 Python/2.7.5 (internal dummy connection)"

error_log :

[Tue Feb 09 18:25:31.704733 2021] [core:notice] [pid 64464] SELinux policy enabled; httpd running as context system_u:system_r:httpd_t:s0 [Tue Feb 09 18:25:31.705529 2021] [suexec:notice] [pid 64464] AH01232: suEXEC mechanism enabled (wrapper: /usr/sbin/suexec) [Tue Feb 09 18:25:31.726489 2021] [lbmethod_heartbeat:notice] [pid 64464] AH02282: No slotmem from mod_heartmonitor [Tue Feb 09 18:25:31.738457 2021] [mpm_prefork:notice] [pid 64464] AH00163: Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips mod_wsgi/3.4 Python/2.7.5 configured -- resuming normal operations [Tue Feb 09 18:25:31.738510 2021] [core:notice] [pid 64464] AH00094: Command line: '/usr/sbin/httpd -D FOREGROUND'

and both

https://slurm-web.localdomain.com/slurm-restapi/cluster https://slurm-web.localdomain.com/slurm-restapi/nodes are working

[rezib]

Hi everyone, thank you all for your interest in Slurm-web!

I just released Slurm-web v2.4.0 included a spec file to build RPM for el8. Please check out!

Let me know if you see any good reason to reopen this issue.