You are here

meshchat v1.01 testing help needed

21 posts / 0 new
Last post
meshchat v1.01 testing help needed


Alphabetical file list
File list column width changes
Fixes for long channel names
An attempt at stretch support. If someone has a pi to try this on that would be great.

To install on a PI:

copy file to your pi
dpkg -i /home/pi/meshchat_1.01_all.deb

You may be prompted if you want to keep your config, in most cases you want to keep your local config. Just check that has what you want in it after install.


Let me know any problems you find. Thanks!


Testing now at http://kd6epq-1-pi1.local.mesh/meshchat

No problems updating. Just had to put the node back in the config. This is a Jesse pi.


nc8q's picture
meshchat v1.01 testing help needed

Fresh install of NOOBS_v_8_2 (because it lets me install / to hard drive).
Change the password for user 'pi'.
Create new user
Give new user the same groups as user 'pi'.
sudo apt-get update
sudo apt-get upgrade
Installed meshchat_1.0_all.ipk on node (Ooopps, fixed later.)
sudo apt-get install curl apache2
dpkg -i meshchat_1.01_all
Edited 'our $pi_zone=' to local usage.
Load meshchat from node into browser and logon.
Send test message:
"Error sending message: Could not get lock"
Logoff of meshchat.
sudo systemctl stop apache2
Edited /lib/systemd/system/apache2.service, set 'PrivateTmp=false'.
sudo systemctl start apache2
 remove meshchat_1.0_all.ipk
 install meshchat_1.01_all.ipk:
  Installing meshchat (1.01) to root...
   Applying service announcement
   Mesh Chat has been setup at http://NC8Q-Loaner-locoM2-167:8080/meshchat
   An advertised service has been added for Mesh Chat on the Services configuration page
   Configuring meshchat.
   Failed to restart all services, please reboot this node.
Rebooted node.
 meshchat_1.01 loads current MESHCHAT messages from local AREDNMESH network.
 Send message works.
/\ fingers crossed.

3s, Chuck

sudo systemctl stop apache2

sudo systemctl stop apache2
Edited /lib/systemd/system/apache2.service, set 'PrivateTmp=false'.
sudo systemctl start apache2

If I read the OP correctly, weren't Stretch users supposed to not have to do that with 1.01?

You should not need to do the

You should not need to do the  'PrivateTmp=false' fix anymore. When the package is installed it checks if its stretch and installs a different config that uses /var/www/html/meshchat vs /tmp

what about meshchat-api?

So if we have a Raspberry Pi installation, do we keep meshchat 1.0 API on the node and upgrade the Pi?


Correct, you only need the 1

Correct, you only need the 1.0 API on the node.

Thanks, Ian.

Thanks, Ian.

I backed out the Stretch

I backed out the Pi Stretch fixes I made awhile back and 1.01 is still working on with meshchat api 1.0
without the previous Stretch fixes. 

Somehow I wound up with a .pm file with our $meshchat_path              = "/tmp/meshchat";
instead of "/var/www/html/meshchat";
and a file I don't remember putting there... but I don't remember a lot these days...

Nice to have the channel field truncated... We're running a local zone, MVMchat, FWIW.

Thanks, Trevor!


k1ky's picture
v1.01 Looking Good

Trevor - Many thanks for this.  I have installed v1.01 all.ipk stand-alone on a Nanostation M2 with AREDN-186.  The "SEND" Button is no longer hidden by my long message list box.  This was an "upgrade" from a 1.0 install. I removed 1.0 first, then re-installed the 1.01 package.  All is well so far.

Looks good

Thanks for the update! I installed on my Raspberry Pi and everything looks good. Fixed the long channel names issue. (Thank you!!) Also Files section column widths look good for long file names! Alphabetical sorting is a nice feature too.

Sorry I don't have a stetch version of Raspbian (running on Wheezy) to try. Thanks again.

stretch testing on 1.01
Hi Trevor,

The test for 9.0 is too restrictive as 9.4 is the current stretch release.

When I modified the /etc/debian_version to 9.0 I get this

Setting up meshchat (1.01) ...
cp: cannot stat '/usr/lib/cgi-bin/': No such file or directory

I saw that there is a type-o in the filenames "mechchatconfig" should be "meshchatconfig" ??


# Check for stretch
if grep -Fxq "9.0" /etc/debian_version
cp /usr/lib/cgi-bin/ /usr/lib/cgi-bin/

Thanks for all your hard work.




Try this with fixes for
stretch testing with 1.02

This worked well:
pi@ki5aax-pi:~ $ sudo dpkg -i /tmp/meshchat_1.02_all.deb
(Reading database ... 106060 files and directories currently installed.)
Preparing to unpack /tmp/meshchat_1.02_all.deb ...
Unpacking meshchat (1.02) over (1.01) ...
Setting up meshchat (1.02) ...
pi@ki5aax-pi:~ $


KM4DC's picture
Works on Buster

Chuck, nc8q,
Thanks for the instructions. I had the "Error sending message: Could not get lock" problem.
Followed your instructions on a RPi 3 with Buster and everything seems good.
Using MeshChat 1.02 on NS M2.
Don, KM4DC

Mesh Chat v1.02 messages disappearing

We've been using Mesh Chat for Davis County, UT for a while now. Running v1.02 on a node. Last week we noticed that many months of messages were gone. There were a bunch of messages sent during our ARES net last week after we discovered the first set of messages gone. A few days after these new messages were posted, they also disappeared. Any suggestions on troubleshooting? It appears from the node page there is memory available on the node (a Nanostation NS2).

free spaceflash = 1416 KB
/tmp = 13928 KB
memory = 3312 KB

Thanks for your help and suggestions.
K7BBR - Brandon

AA7AU's picture
You mentioned that you have

You mentioned that you have it installed "on a node". You probably already know this, but the entire MeshChat textfile of messages (database) is based in memory which gets reset when a node is power-cycled. However, if another node, connected by RF, has the exact same zonename it will have an automatically sync'd copy and your node will repopulate its database from that shortly after it powers back up and connects by RF. That database (typos and all) will survive as long as at least one connected zonename node stays powered up. If all instances are powered down, it will be lost. Some use Raspberry Pi installs to overcome this for single node installs.

- Don - AA7AU  (who generally has several instances running concurrently just to be safe and archived)

Don - Thanks for the reply.

Don - Thanks for the reply. Makes sense now that the memory doesn't persist when the node is reset. I didn't realize that about Mesh Chat and the node memory. In the past we had as many as 5 instances of Mesh Chat on 5 nodes plus a Raspberry Pi. That seemed to be too many for our instance, in that the nodes would talk back and forth and slow things down. Looks like at least 2 instances would be good so we have message backup. We'll get another instance spooled up. It's a great program and tool for our group. Thank you!
Brandon - K7BBR

AE6XE's picture
/tmp available space is

/tmp available space is misleading...

/tmp = 13928 KB
memory = 3312 KB

The /tmp file system space where meshchat messages are stored is defined to be a fixed and given size.   Of this size, we see how much space is remaining.   However, since /tmp space is also physcially in RAM, it additionally is limited by how much RAM is available.   In this case with 3.3M of RAM remaining, the node has limited space to function.   

All the functions of the device including ,meshchat messages, routing, responding to UI requests, and more are using this limited RAM.     One possible explanation is during a busy point in time, /tmp could not allocate space and some of the messages were lost (and there wasn't another meshchat instance to sync back with). 


Makes sense about the limited

Makes sense about the limited memory. Could have been the limited RAM and high node usage. Would explain things. Also looks like the location where the node is installed at our city hall had a power bump and the node reset a few days ago.

Thanks for your reply and for your hard work on the AREDN firmware!
Brandon - K7BBR

AA7AU's picture
The RF "chatter" between

The RF "chatter" between different instances of Meshchat using the same zonename is basically swapping a hash-code based on database size. Way less than OLSR chatter. If no messages have been added (individual messages cannot be deleted), there is no additional "synch" traffic. When you bring up a node which has been powered down, it may take a minute or two for it to get a new copy of the database. But remember that is simply text and not terribly large, less than the size of a very small image (a year of ours is less than 50k of text but it's not terribly active). Download your own copy and look at it. Meshchat defines Total Storage: 512.0 KB

Of course, I defer to Joe on the technicallities on all this - his advice is always spot on.

- Don - AA7AU

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer