Large data consumption

[mahdi1234]

I was comparing Simplex with some other SW with central server and frankly I found idle data consumption rather high.

All my tests were done moving in country side where signal strength varied between LTE and Edge. Pure LOS without gapps so all apps using background service. Results below are after 90 minutes of these tests:

• Element (subscribed to 60 rooms) - 154kB
• Schildichat (subscribed to few small rooms) - 88kB
• Signal (few contacts, couple of groups) - 141 kB
• Simplex (two contacts, no activity at all) - 5.60 MB

I would expect with no notifications/messages sent to be much closer to other apps.

[epoberezkin]

Could you please post chat console log from these tests?

The data consumption very much depends on the connection quality, and if the app keeps trying to subscribe to the queues and can't do it (it starts from retrying every second, and then backs of to every 10 seconds) - e.g. because of the timeouts - it would keep sending 16kb messages, 1 per contact.

We probably should have slightly less aggressive settings for mobile devices - e.g. longer timeout and retry intervals.

[mahdi1234]

Is there a way how to export it from console? Seems I don't have that older logs anymore, so will try to repeat next weekend.

Please note all four apps were running in parallel so carrier situation was identical for all of them.

Is there a way how to export it from console? Seems I don't have that older logs anymore, so will try to repeat next weekend.

Please note all four apps were running in parallel so carrier situation was identical for all of them.

Aye, did u retry? 🙂

[epoberezkin]

Is there a way how to export it from console?

sorry, no you can't access old logs - they only exist while the app is running.

[mahdi1234]

Ok, but how do I export it? I can see log, but don't see way how to get that data out.

[polyrod]

Hi, its the second month simplex used up all my 3gb mobile data transfer volume, i looked at the console last month and saw hundrets of bytes polling every 1 or 10 seconds , thats way too much to check if new messages are in queue. This has to be resolved. If those 100 byte polls are redundant , i.e they check for the samr stuff , couldn't the request be saved once serverside and and triggered by a client poll request with some short request key ? So instead of transmittting 100s of bytes every second only transmit 10 or so ...

[epoberezkin]

Thanks for the info... Interestingly, we do not poll for the messages – we only have an open socket and the server is supposed to push new messages when they are available. In addition to that the client sends ping every 10min and receives the response. These pings should consume up to ~200Mb a month (or less, if part of the time the device is offline – assuming 3 default servers, the calculation is 16kb * 6 times/hour * 24hours * 30 days * 3 servers).

So what might be happening here is one of two things:

1. the device is losing connection and reconnecting, but it would not be 100s of bytes - it would be at least some kilobytes,
2. TCP keep-alive traffic - it is configured to send a small packet every 15 seconds to keep the socket open.

Given that this doesn't seem to affect all users, I am thinking that maybe different devices/OSs do it differently?

My suggestion would be that we add some advanced network configuration to be able to enable/disable TCP keep alive entirely and to optionally change its frequency and also do the same for pings.

Thanks a lot for the patience and for feedback, we are absolutely not going to ignore this problem. It's a full time project for us, and making it stable and performant, particularly in terms of battery and traffic consumption, is a high priority for us - what you experience should not be a norm.

I'll ping here once we release these advanced settings - it probably won't be in 3.1 due to release on Monday/Tuesday, but probably in the next 3.1.1 (or 3.2) release a week later – it would be great if you could experiment with these settings to see how they affect your traffic, without giving it a month, just a few days should show the difference...

Thank you!

[polyrod]

I ve deinstalled simplex for now , and booked 1 extra gb , i will reinstall when i get my next 3gb and then i will check and can also post logs n stuff, i am on android . Thanx for your fast reply. Evgeny Poberezkin ***@***.***> schrieb am Mi., 27. Juli 2022, 15:36:

…

Thanks for the info... Interestingly, we do not poll for the messages – we only have an open socket and the server is supposed to push new messages when they are available. In addition to that the client sends ping every 10min and receives the response. These pings should consume up to ~200Mb a month (or less, if part of the time the device is offline – assuming 3 default servers, the calculation is 16kb * 6 times/hour * 24hours * 30 days * 3 servers). So what might be happening here is one of two things: 1. the device is losing connection and reconnecting, but it would not be 100s of bytes - it would be at least some kilobytes, 2. TCP keep-alive traffic - it is configured to send a small packet every 15 seconds to keep the socket open. Given that this doesn't seem to affect all users, I am thinking that maybe different devices/OSs do it differently? My suggestion would be that we add some advanced network configuration to be able to enable/disable TCP keep alive entirely and to optionally change its frequency and also do the same for pings. Thanks a lot for the patience and for feedback, we are absolutely not going to ignore this problem. It's a full time project for us, and making it stable and performant, particularly in terms of battery and traffic consumption, is a high priority for us - what you experience should not be a norm. I'll ping here once we release these advanced settings - it probably won't be in 3.1 due to release on Monday/Tuesday, but probably in the next 3.1.1 (or 3.2) release a week later – it would be great if you could experiment with these settings to see how they affect your traffic, without giving it a month, just a few days should show the difference... Thank you! — Reply to this email directly, view it on GitHub <#688 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AF5ZZYQ2QFBURGP2473YDJLVWE3MHANCNFSM5WTZ2HOQ> . You are receiving this because you commented.Message ID: ***@***.***>

[epoberezkin]

Yep, and let me do the advanced settings. I think either disabling keep-alive can solve this problem or I have some other ideas where to improve how we do networking - it's definitely not message polling, it's some lower level problem :)

The reason we enabled keep-alive ini the first place is because we observed such behaviours when the client thought it's still connected to the server, when in fact it already lost the connection, and the opposite, and sending more frequent protocol-level pings is more expensive...

[epoberezkin]

@polyrod btw, we just discussed several possibilities here... when you say 100s of bytes - are you sure it's 100s of bytes rather than 16kb block? where do you monitor it?

Any logs would be helpful - could you maybe send via chat ("connect to developers" option) or to chat@simplex.chat?

[epoberezkin]

@polyrod advanced network settings are released, but before you test too much - have you (or your contact) by any chance tried using the messaging server you host yourself?

I've just understood what is the scenario when the app will generate approximately that amount of traffic... I wish I did before doing these network settings :)

[polyrod]

When i checked , i have used the chat console to see what messages where exchanged between app and server, there i saw it was always the same "conversation" , that where ~10 lines (/command params) with each line approx 10 chars iirc. The timestamps suggested it was repeating once every second or once every 10 seconds , i am unsure in retrospect. I get new transfer volume at the 12th of this month, then i will reinstall simplex app , and can check and send you logs. Evgeny Poberezkin ***@***.***> schrieb am So., 7. Aug. 2022, 18:30:

…

@polyrod <https://github.com/polyrod> advanced network settings are released, but before you test too much - have you (or your contact) by any chance tried using the messaging server you host yourself? I've just understood what is the scenario when the app will generate approximately that amount of traffic... I wish I did before doing these network settings :) — Reply to this email directly, view it on GitHub <#688 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AF5ZZYXNBXJPZCERUC2B5UDVX7QBXANCNFSM5WTZ2HOQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[epoberezkin]

great - thank you!

What I meant is whether you or your contacts used your own servers that are no longer running - that may be the reason for the traffic... There will be a fix to that problem soon, but it's not in the current version.

[polyrod]

No , we haven't setup our own servers , we used the ones provided by the app. Evgeny Poberezkin ***@***.***> schrieb am So., 7. Aug. 2022, 19:29:

…

great - thank you! What I meant is whether you or your contacts used your own servers that are no longer running - that may be the reason for the traffic... There will be a fix to that problem soon, but it's not in the current version. — Reply to this email directly, view it on GitHub <#688 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AF5ZZYV2DCYF4SBLTLX4JZ3VX7W7PANCNFSM5WTZ2HOQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[epoberezkin]

We’ve just identified a scenario that would cause similar amount of traffic, and the fix for it is coming - if you have the message that the app cannot delivere it would keep trying, and it would create the traffic. It would be great to confirm if you the same issue or is it something else. Would you be able to export the chat database, copy to the laptop and make several queries I would send here?