We have been experiencing a significant issue in our installation which is going to be a deal-breaker here soon if we cannot get this resolved.
NSD keeps taking up all of the available memory and CPU and will not release. Killing the process and allowing it to respawn helps for a short time, but not enough to allow all our users to effectively utilize the system.
Here are the stats that you will require:
ProjectOpen 4.0.3.1.0 - latest build and just ran the ./all-packages.sh update + relevant update scripts yesterday
VMware ISO on VMware ESXi 4.0
Using the VMware Template for RedHat Linux 5 32-bit
PostgreSQL v 8.4.11
2 CPUs, 4GB RAM
My thought is that I would either need to rebuild this system from scratch or completely abandon this project, neither of which I would prefer to do because of all the data we would lose.
Any help would be appreciated and we are willing to do what is necessary to get this system back up and functional.
Thank you so much!
J.R.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I know there are some memory leaks in certain versions of aolserver. Which is why we go with Ubuntu 10.4 lts and run it with apt-get.
Additionally we experienced issues with VMware hosted virtual machines, which drove us to use KVM and cloudmin ( tried Xeon as well, didn't really work that great). With VMware we restarted the server once per night and all was swell when we did our backups. Now the uptime is as long ago as when we setup the machine (fingers crossed).
If you have a lot of users, consider using NGINX as your reverse proxy. It does a great job and you could configure it to serve eg. The files directly instead of taking up valuable aolserver resources,
Make sure your provide ample of ram to Postgres. I know this sounds counter intuitive, but with a lot of ram the DB gets faster which means queries are shorter, frees up aolserver waiting time… You get the picture.
I hope that helps you go in the right direction. I have some more tricks up my sleeve but that would be consulting (as it contains the need to possibly change code pieces) so I would first give it a try wi Ubuntu 10.4 and follow my instructions on the Cognovis website.
Best wishes
Malte
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
We checked with our current ]po[ V4.1 development server (CentOS on ESXi) and there is plainly no such issue. Also, this same configuration is running literally at thousands of customers, so any general issue would have been reported a lot more frequently.
Is there any difference between your installation and the vanilla VMware image? Did you write custom code? Did you enable less frequently used features in ]po[? How many currently active users are there?
We do offer to enter into your machine and look into the issue in detail, but only as part of a service contract. Sorry, we somehow need to earn our money, if the software is free ...
Cheers!
Frank
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello!
We have been experiencing a significant issue in our installation which is going to be a deal-breaker here soon if we cannot get this resolved.
NSD keeps taking up all of the available memory and CPU and will not release. Killing the process and allowing it to respawn helps for a short time, but not enough to allow all our users to effectively utilize the system.
Here are the stats that you will require:
ProjectOpen 4.0.3.1.0 - latest build and just ran the ./all-packages.sh update + relevant update scripts yesterday
VMware ISO on VMware ESXi 4.0
Using the VMware Template for RedHat Linux 5 32-bit
PostgreSQL v 8.4.11
2 CPUs, 4GB RAM
An excerpt from ps:
root 27149 0.0 0.0 6044 2180 ? Ss 11:10 0:00 su -c /usr/local/aolserver451/bin/nsd -it /web/projop/etc/config.tcl -u projop -g projop projop
projop 27159 15.9 8.7 384484 349884 ? Ssl 11:10 5:39 /usr/local/aolserver451/bin/nsd -it /web/projop/etc/config.tcl -u projop -g projop
My thought is that I would either need to rebuild this system from scratch or completely abandon this project, neither of which I would prefer to do because of all the data we would lose.
Any help would be appreciated and we are willing to do what is necessary to get this system back up and functional.
Thank you so much!
J.R.
Hi jbullington. I am getting the exact same problem right now. Did you ever solved the problem?
I know there are some memory leaks in certain versions of aolserver. Which is why we go with Ubuntu 10.4 lts and run it with apt-get.
Additionally we experienced issues with VMware hosted virtual machines, which drove us to use KVM and cloudmin ( tried Xeon as well, didn't really work that great). With VMware we restarted the server once per night and all was swell when we did our backups. Now the uptime is as long ago as when we setup the machine (fingers crossed).
If you have a lot of users, consider using NGINX as your reverse proxy. It does a great job and you could configure it to serve eg. The files directly instead of taking up valuable aolserver resources,
Make sure your provide ample of ram to Postgres. I know this sounds counter intuitive, but with a lot of ram the DB gets faster which means queries are shorter, frees up aolserver waiting time… You get the picture.
I hope that helps you go in the right direction. I have some more tricks up my sleeve but that would be consulting (as it contains the need to possibly change code pieces) so I would first give it a try wi Ubuntu 10.4 and follow my instructions on the Cognovis website.
Best wishes
Malte
Hi Fernando,
This seems to be the VMware appliance from ]po[ V3.5.
I just can say that we are running ]po[ V4.0 (aolserver451rc2 with TCL version 8.5) with an uptime of 500 days on one of our production servers.
However, we do perform a "killall -9 nsd" every days at 5am in the morning.
The AOLserver source code is available here on SourceForge at:
https://sourceforge.net/projects/project-open/files/project-open/Support%20Files/
Cheers!
Frank
Hi, i'm getting the same memory problems:
CentOS 5.3
AOL Server 4.5.1
]PO[ 4.0.5
Running on VMWare desktop or VMWare ESXI the problem is the same, does anyone have a solution for this?
Tks!
Last edit: António Torres 2015-02-26
Hi António,
We checked with our current ]po[ V4.1 development server (CentOS on ESXi) and there is plainly no such issue. Also, this same configuration is running literally at thousands of customers, so any general issue would have been reported a lot more frequently.
Is there any difference between your installation and the vanilla VMware image? Did you write custom code? Did you enable less frequently used features in ]po[? How many currently active users are there?
We do offer to enter into your machine and look into the issue in detail, but only as part of a service contract. Sorry, we somehow need to earn our money, if the software is free ...
Cheers!
Frank
Hi Frank,
can you, please tell me more about the service contract?
Thanks!
Hi!
http://www.project-open.com/en/services/project-open-support.html
Because our software is free, we need to make money with
the services around it in order to pay the bills of our
core staff.
We do the support here on SourceForge as good as we can,
but we can't really look into customer's systems.
Cheers!
Frank
Hey Frank no Prob! I understand that, thanks a lot! :)