[Проблема] Проблема на co-location

[Dolphin]

Сообщение от Denis Shlyapnikov

Ну что там, ребята? На почту никто не отвечает. В отправленные контакты никто не пишет. = ( А ситуация оставляет желать лучшего.

Shark Telecom:

64 bytes from 91.212.89.137: icmp_req=190 ttl=58 time=63.2 ms
64 bytes from 91.212.89.137: icmp_req=191 ttl=58 time=60.3 ms
64 bytes from 91.212.89.137: icmp_req=192 ttl=58 time=57.5 ms
64 bytes from 91.212.89.137: icmp_req=193 ttl=58 time=54.8 ms
64 bytes from 91.212.89.137: icmp_req=194 ttl=58 time=62.0 ms
64 bytes from 91.212.89.137: icmp_req=195 ttl=58 time=62.7 ms
64 bytes from 91.212.89.137: icmp_req=196 ttl=58 time=59.5 ms
64 bytes from 91.212.89.137: icmp_req=197 ttl=58 time=60.7 ms
64 bytes from 91.212.89.137: icmp_req=198 ttl=58 time=61.9 ms
64 bytes from 91.212.89.137: icmp_req=199 ttl=58 time=64.2 ms
64 bytes from 91.212.89.137: icmp_req=200 ttl=58 time=50.9 ms

--- 91.212.89.137 ping statistics ---
200 packets transmitted, 191 received, 4% packet loss, time 199285ms
rtt min/avg/max/mdev = 45.633/59.297/182.025/14.141 ms

Uzinfocom DC (Машина на коллокейшене, скорее всего общий свич с вашей):

64 bytes from 91.212.89.137: icmp_req=190 ttl=63 time=21.0 ms
64 bytes from 91.212.89.137: icmp_req=191 ttl=63 time=20.3 ms
64 bytes from 91.212.89.137: icmp_req=192 ttl=63 time=18.3 ms
64 bytes from 91.212.89.137: icmp_req=193 ttl=63 time=17.8 ms
64 bytes from 91.212.89.137: icmp_req=194 ttl=63 time=18.3 ms
64 bytes from 91.212.89.137: icmp_req=195 ttl=63 time=27.2 ms
64 bytes from 91.212.89.137: icmp_req=196 ttl=63 time=9.77 ms
64 bytes from 91.212.89.137: icmp_req=197 ttl=63 time=15.0 ms
64 bytes from 91.212.89.137: icmp_req=198 ttl=63 time=19.0 ms
64 bytes from 91.212.89.137: icmp_req=199 ttl=63 time=10.2 ms

--- 91.212.89.137 ping statistics ---
200 packets transmitted, 180 received, 10% packet loss, time 199311ms
rtt min/avg/max/mdev = 5.991/17.630/29.200/5.866 ms

Для сравнения пинг до вышеуказанного сервера на том же коллокейшене (обычное железо с абу-сахи, средний трафик - гигабайт в час):

64 bytes from 91.212.89.70: icmp_req=190 ttl=58 time=37.0 ms
64 bytes from 91.212.89.70: icmp_req=191 ttl=58 time=39.2 ms
64 bytes from 91.212.89.70: icmp_req=192 ttl=58 time=40.7 ms
64 bytes from 91.212.89.70: icmp_req=193 ttl=58 time=39.2 ms
64 bytes from 91.212.89.70: icmp_req=194 ttl=58 time=40.4 ms
64 bytes from 91.212.89.70: icmp_req=195 ttl=58 time=39.0 ms
64 bytes from 91.212.89.70: icmp_req=196 ttl=58 time=36.7 ms
64 bytes from 91.212.89.70: icmp_req=197 ttl=58 time=38.4 ms
64 bytes from 91.212.89.70: icmp_req=198 ttl=58 time=40.2 ms
64 bytes from 91.212.89.70: icmp_req=199 ttl=58 time=39.2 ms
64 bytes from 91.212.89.70: icmp_req=200 ttl=58 time=40.4 ms

--- 91.212.89.70 ping statistics ---
200 packets transmitted, 200 received, 0% packet loss, time 199275ms
rtt min/avg/max/mdev = 36.743/41.671/224.673/19.555 ms

,
и его htop:
.


Вывод - роутер датацентра и ширина сиськи тасикса не виноваты. Чем меньше задержка пакетов (следовательно, больше нагрузка), тем быстрей ваша машина захлебывается и начинает их дропать.
Смотрите железо и конфигурацию, проблемы на вашей стороне.