Score:0

system load spikes causing php-fpm pools to fill and cause micro outages

cn flag

kind of at my wits end at this point and looking for some help.

2 physical Web servers

CentOS NGINX PHP5 php-fpm

2 physical DB Servers

CentOS NGINX elastic search zookeeper couchbase

as user load increases and the php-fpm pools fill I start to get datadog alerts that the network connections are dropping. the odd thing is that I dont see any dropped packets. During the periods that user load increases I have ping floods running across the LAN from server to server from server to outside and from outside to WAN IPs of the servers - and dont get any dropped packets during these time periods

However, when the load is high during the micro outages, commands over ssh, including ls and vim are slow to respond, but don't fail. I can bring up htop which doesn't show anything other than the load values higher than normal at the time.

Im working on setting up a zabbix server to try and catch some of these outages but Im wondering/hoping other people have some suggestions on how to catch these micro outages

jp flag
Dropped packets and dropped connections are different things.
John Bell avatar
cn flag
@AlexD Hi Alex, yeah I understand that, but this is the information I have to go on, if theres something else that you think would help you to suggest some pointers Id be happy to provide it
I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.