Hello everyone, Khabrovites!
Actually, I was prompted to write this article by the frequently triggered triggers of the Zabbix monitoring system for a speed drop in the Tele2 network. At remote sites, to which it is impossible to conduct optics, port forwarding of the registrar is organized through USB modems, reflashed for any operators. Here is a simple equipment connection diagram:
By the way, our organization simultaneously uses the Megafon mobile Internet for the same needs, but there are no such problems with it. And the problem is the following: starting around lunchtime on 07.08.2019/XNUMX/XNUMX, Zabbix notification about the zero speed of the web script execution on the pages of video recorders became so frequent that the sender's mail specified in the Zabbix notification began to perceive trigger emails as spam and not deliver them to the recipient, that is, me. It is also important that the background speed of problematic networks as a whole satisfies, and the drops, although short-term, immediately drop to zero. And if the registrar channels are βwrittenβ to remote media, then problems with the integrity of the recording will inevitably arise, at least, or even completely with the impossibility of reconnecting to the target registrar.
At the time of this writing (19/25/10.08.2019 2:2), the speed drop frequency began to appear less frequently, however, if a drop occurs, then these are the recorders forwarded through Tele2. Here it is very important to note that we are talking about the whole of Moscow and the Moscow region. Even increasing the period between checks to 2 minutes did not practically change the situation. Also, returning the default zabbix_server.conf did not help at all (well, you never know). A call to the Tele2 hotline did not bring any explanation of what was happening, except for the advice to try to set the mixed mode in the modem parameters (apparently it was about βLTE + UMTSβ), which was not required before, and if it was required, then not at all objects with operator "Tele2" at the same time, and the problem clearly lies deeper than the choice of protocol. And I, of course, will contact the corporate department of TeleXNUMX on Monday to clarify the nature of this behavior of the network, but I strongly doubt that THERE will want to somehow discredit themselves. Therefore, this paste, as it were, warns, and at the same time asks: βHave anyone noticed this in their mobile Internetβ TeleXNUMX βor Zabbix?
The folder for receiving Zabbix notifications is in such a s@sr@n state (and so on for 4 days in a row):
Zabbix map with video recorders (if it is not clear in the mail screen that the operator is "Tele2", then it is obvious on the map):
Well, the icing on the cake for all the previously given facts is the synchronism of falls in Moscow and the Moscow region, regardless of their geography. This does not include very distant points like Dolmatovo and Ferma, although, just look at how toothy their charts look:
Here are the fall times in Sgonniki:
Here is the location of the Scoundrels:
Around the same time, the fall in Strogino:
Location of Strogino:
And finally, the same "zero" in Shcherbinka:
She is here:
A small time out of sync is due to the difference at the beginning of the check on the Zabbix server, but in general, as we can see, the network behavior on the graphs is identical at the time of this particular check. Thank you for your attention, I will wait from you at least some similar observations in terms of those related. All for now.
Source: habr.com