Hello everyone, Habr readers!
In fact, the frequent triggers from the Zabbix monitoring system regarding speed drops in the "Tele2" network prompted me to write this article. On remote locations where fiber optics cannot be installed, port forwarding for the recorder is organized through USB modems reprogrammed for various operators. Here is a simple diagram of the equipment connection:
By the way, our organization also uses mobile internet from "Megafon" for the same needs, but no such issues have been observed with it. The problem is as follows: starting approximately from noon on 07.08.2019, the Zabbix notifications about zero speed execution of web scenarios on the video recorder pages became so frequent that the sender's email address indicated in the Zabbix notification began to treat the trigger emails as spam and did not deliver them to the recipient, namely me. It is also important to note that the background speed of the problematic networks is generally satisfactory, and while the drops are brief, they go directly to zero. If the recorders' channels are "writing" to remote storage, there will inevitably be issues with the integrity of the recording at the very least, or even with the impossibility of reconnecting to the target recorder.
At the time of writing this article (19:25 10.08.2019), the frequency of speed drops has decreased. However, when drops do occur, they specifically affect the routers connected through 'Tele2'. It is important to note that this applies to all of Moscow and the Moscow region. Even extending the interval between checks to 2 minutes did not significantly change the situation. Additionally, reverting to the default zabbix_server.conf did not help (just in case). A call to the 'Tele2' hotline provided no explanations for what was happening, other than the suggestion to set the modem parameters to mixed mode (presumably referring to 'LTE+UMTS'), which had not been required before, and if it had been, it was not necessary for all locations with the 'Tele2' operator at the same time. The issue seems to stem from deeper problems than simply choosing a protocol. I will, of course, contact the corporate department of 'Tele2' on Monday to clarify the nature of this network behavior, but I seriously doubt they will want to discredit themselves. Thus, this post serves as a warning and also inquires: "Has anyone noticed something similar in their 'Tele2' mobile internet or in Zabbix?"
The folder for receiving Zabbix notifications is in such a dismal state (and has been for 4 consecutive days):
Zabbix map with video recorders (if it's unclear in the email screenshot that the operator is 'Tele2', it is obvious on the map):
And the cherry on top of all the previously mentioned facts is the synchronicity of drops in Moscow and the nearby Moscow region, regardless of their geography. Very distant points like Dolmatovo and Ferma are not included here, but just take a look at how sharply their graphs appear:
Here's the drop time in Sgoniki:
Here is the area where Sgoniki are located:
Around the same time, there was a drop in Strogino:
Location of Strogino:
And finally the same 'zero' in Shcherbinka:
It is located here:
A slight time desynchronization is due to the difference in the start of the checks on server Zabbix, but overall, as we can see, the network behavior on the graphs is identical during this specific check. Thank you for your attention; I will be waiting for any similar observations from you regarding this matter. Goodbye to everyone.
Source: habr.com
