OpenLiteSpeed with DirectAdmin

1. how can i access to admin web console via port 7080
2. lscache not working

1. Enable WebaAmin Console
- Enable WebAdmin console in /usr/local/lsws/conf/httpd-defaults.conf. Just change disableWebAdmin 1 to 0
- Create symlink ln -sf /usr/local/lsws/admin/fcgi-bin/admin_php5 /usr/local/lsws/admin/fcgi-bin/admin_php
ln -sf /usr/local/lsws/admin/html.open /usr/local/lsws/admin/html
- Change WebAdmin Port in /usr/local/lsws/admin/conf/admin_config.conf
- Create Log Directory mkdir -p /usr/local/lsws/admin/logs
- Create Admin Pass: sh /usr/local/lsws/admin/misc/admpass.sh
2. Fix LSCache
chow apache.apache /usr/local/lsws/cachedata/
 
Per client throttle doesn’t work on directadmin’s openlitespeed for some reason.
 
I had and issue with OLS , For some strange reason every nigh 2:20 am it start to spawn process uncontrolled and deleting core dumps uncontrolled until the VM got absurd levels of load

systemctl restart litespeed stop that but it is strange as it happens in same hours for the 3 days of tests

I currently run the latest build since yesterday

./directadmin o
Compiled on 'CentOS 7.0 64-Bit'
Compile time: Jul 7 2019 at 15:42:44
Timestamp: '1562535715'
Compiled with IPv6

It also happens on the old build from June 27 that's why I decide to go with latest

https://ibb.co/LQx3rwn

process.png



I do not know if its related but I setup cron to do daily update as this is test VM on which I'll try to break DA in many various ways before i use it to host websites

#Cronjob Settings
cron=yes
cron_frequency=daily
email=admin@XXXXX
notifications=yes
da_autoupdate=yes
updates=yes
webapps_updates=yes


today happens during the night at 2:20 sever time (sever is in EDT time zone) and I noticed it again during the day so maybe it is different issue

Does any one get any idea
 
OLS console on :7080 is enabled by default in CB 2.0 rev. 2159, it's read-only mode. It'd be great if you may report issues/bugs detected in WebAdmin should to OLS developers (they have Slack channel for it) :) Admin pass generated will be restored at the end of "./build openlitespeed", stored in /usr/local/directadmin/scripts/setup.txt (if pass is updated - it won't be changed there). Thank you.
 
Heavy memory consumption

Is any other experiencing heavy memory comsumption (also CPU) with OpenLiteSpeed + Cloudlinux + Directadmin?

50 concurrent users (loadimpact.com service) on a wordpress new installation consumes 8-16GB memory and a high load. Something is really wrong...

[root@server ~]# /usr/local/lsws/bin/lshttpd -v
LiteSpeed/1.4.48 Open (built: Mon Jun 24 21:47:57 UTC 2019)
module versions:
modpagespeed 2.2-1.11.33.4
cache 1.61
modinspector 1.1
uploadprogress 1.1
mod_security 1.1
 
Don't see anything wrong on my side:

Code:
[root@server1 custom]# /usr/local/lsws/bin/lshttpd -vLiteSpeed/1.4.48 Open (built: Mon Jun 24 21:47:57 UTC 2019)
        module versions:
        modpagespeed 2.2-1.11.33.4
        cache 1.61
        modinspector 1.1
        uploadprogress 1.1
        mod_security 1.1


[root@server1 custom]#

and

Code:
[root@server2 ~]# /usr/local/lsws/bin/lshttpd -v
LiteSpeed/1.5.2 Open
        module versions:
        modpagespeed 2.2-1.11.33.4
        cache 1.61
        modinspector 1.1
        uploadprogress LSIAPI_VERSION_STRING
        mod_security 1.1
 (built: Sat Jul  6 12:43:26 UTC 2019) (DEBUG)
        module versions:
        modpagespeed 2.2-1.11.33.4
        cache 1.61
        modinspector 1.1
        uploadprogress LSIAPI_VERSION_STRING
        mod_security 1.1


[root@server2 ~]#


without CloudLinux though....
 
There is a bug somewhere in CL/OLS

I migrated a site from a cPanel shared hosting, running CL + LS. HIstorical data was showing 80MB-120MB for that particular wordpress site.
After migration to DA + CL + OLS, it was consuming 1GB memory. So almost 10x more.

Now i cant figure out if its related to CL or OLS or DA. Tried to talk with DA guys, but seems like the BUG is somewhere in OLS/CL.

Is anyone running DA + CL + OLS/LS ? You may not be aware of that high usage. If you are running DA+OLS only, have a look on memory usage for a particular user.
There must be other who are having this problem.
 
There is a bug somewhere in CL/OLS

I migrated a site from a cPanel shared hosting, running CL + LS. HIstorical data was showing 80MB-120MB for that particular wordpress site.
After migration to DA + CL + OLS, it was consuming 1GB memory. So almost 10x more.

Now i cant figure out if its related to CL or OLS or DA. Tried to talk with DA guys, but seems like the BUG is somewhere in OLS/CL.

Is anyone running DA + CL + OLS/LS ? You may not be aware of that high usage. If you are running DA+OLS only, have a look on memory usage for a particular user.
There must be other who are having this problem.

same as what I experienced, high usage at night even though there is no active user. finally re-activate apache and nothing like that ever happens again.
 
same as what I experienced, high usage at night even though there is no active user. finally re-activate apache and nothing like that ever happens again.

Have you try Reverse-proxy NGINX + Apache and mod_lsapi?

I also convert my backup from CP to directadmin and all sites give 404 error tested on nginx and openlitespeed
 
Have you try Reverse-proxy NGINX + Apache and mod_lsapi?

we are using this configuration right now, its working well. try litespeed to get a comparison of Apache Nginx, only the problem is the high server load suddenly ;)
 
its mean no problem anymore with cloudlinux also ?

I made some tests and conclusion is that OpenLiteSpeed (whatever version) has a bug which makes the load 10x more than normal. Tried all version like 1.4.48, 1.5.2, 1.5.3. For sure this issue is present in CL+DA+OLS environment. I don't know if non-CL is affected or not. According to zEIter it doesnt seem to be an issue.
1.5.3 had an improvement, but still not close to "normal"
... I installed the enterprise trial version, and it dropped immediately. So problem can be isolated to OpenLiteSpeed.

I am in contact with OpenLiteSpeed guys and DirectAdmin team as well. OLS team will have a look on monday.
Nothing can beat litespeed in performance. Spend weeks on making benchmarks, and litespeed win (no difference on OpenLiteSpeed and LiteSpeed Enterprise performance)
 
Last edited:
Tested 2 servers with CentOS 7 and DirectAdmin. A default WP was used for testing purposes, the latest version.

1. 4GB RAM + 1 core hosted at intovps.com (they offer free DirectAdmin by the way): OpenLiteSpeed + lsphp
2. 4GB RAM + 2 cores hosted at transip.eu: Nginx+Apache+PHP-FPM

Used wrk for testing purposes.

100% CPU usage can be seen on the both servers with 5+ requests/sec. Tried running tests for 30-300 seconds.

- OpenLiteSpeed + lsphp (highest 1min LA noticed at ~17.00)

Code:
Running 30s test @ https://example.net/
  1 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.63s   107.11ms   1.98s    73.94%
    Req/Sec    14.11     17.68    80.00     92.42%
  172 requests in 30.03s, 44.51KB read
  Socket errors: connect 0, read 0, write 0, timeout 7
Requests/sec:      5.73
Transfer/sec:      1.48KB



Running 5m test @ https://example.net/
  1 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.65s    96.60ms   2.00s    77.40%
    Req/Sec     8.20      8.22    90.00     90.91%
  1789 requests in 5.00m, 462.97KB read
  Socket errors: connect 0, read 0, write 0, timeout 68
Requests/sec:      5.96
Transfer/sec:      1.54KB


The both servers do not trouble me during other time.
 
I made some tests and conclusion is that OpenLiteSpeed (whatever version) has a bug which makes the load 10x more than normal. Tried all version like 1.4.48, 1.5.2, 1.5.3. For sure this issue is present in CL+DA+OLS environment. I don't know if non-CL is affected or not. According to zEIter it doesnt seem to be an issue.
1.5.3 had an improvement, but still not close to "normal"
... I installed the enterprise trial version, and it dropped immediately. So problem can be isolated to OpenLiteSpeed.

I am in contact with OpenLiteSpeed guys and DirectAdmin team as well. OLS team will have a look on monday.
Nothing can beat litespeed in performance. Spend weeks on making benchmarks, and litespeed win (no difference on OpenLiteSpeed and LiteSpeed Enterprise performance)

Yes you are right, i also test and it feel OLS + DA + CL combo put EXTRA load on server
 
Back
Top