17

So, the problem here turned out to be a hardware issue. Things were working fine a month ago, and one does not expect failed hardware to still "work" in a degraded mode, but that's what was happening. The troubleshooting step that revealed the issue was to actually look at the ethernet port lights on the back of the cable modem. Instead of the green "...


7

A fairly common infrastructure is one where none of the actual application servers have public IPv4 IP-addresses, they will be in a RFC 1918 private network range behind a load-balancer and any outgoing request they make is either: routed through a NAT gateway, giving the appearance of a single source IP-address must be made through a proxy server, which ...


7

What I would do here is revert the changes individually and run a speed test after each or revert all the changes. Benchmark an unmodified version of CentOS (baseline) and then apply each change individually and run the speed test after each change.


5

The flexibility of IPv6 with addressing is great, but it does indeed make things like this harder. An algorithm that I would recommend: Start by blocking separate IPv6 addresses (a /128). It might be a single user on a network with multiple users, and you want to avoid blocking innocent users (which happens with IPv4 all the time because of NAT, let's not ...


4

Use a status code other than 503 for your "maintenance mode." As we can clearly see, users don't actually get served a 503 when you are using "maintenance mode" anyway, so there's no benefit to using that status code internally in your configuration. Make up another code (593?) and use that. Or better yet, skip the extra location and just send the rewrite ...


4

If you are using google chrome or chrome based browser right click anywhere in the page and click inspect element or (Control + Shift + I) if on windows, the click the "network" tab, change the no throttling option to the speed of your choice... Am sure you will be fine as long as the DevTools are open. you will find similar option on other browsers.


4

Postfix is not being particularly aggressive. It's trying to deliver mail in accordance with SMTP and various best practices. This is Yahoo's fault. Yahoo has extremely low limits on simultaneous recipients of a message from a single sender. I see this occur with a small 500-member list with only a couple of dozen Yahoo addresses on it. The contents of ...


3

dont have a machine running IIS ARR infront of me at the moment, but found the following: Server Farms > FarmName > Proxy > Buffer section > Response Buffer Threshold Via here. Hope this helps...


2

I thought I would post a update since the project is now completed successfully using a NAT instance. Is using "a NAT instance" something I should be looking into? It looks complicated - Is there not a simpler solution? Now that the NAT instance is all set up, the initial feeling of complexity has passed and it actually feels quite simple - even cleaner ...


2

So the issue, which I learned from the answer to this question, is that kernel buffering bypasses the blkio system. Using fio with no buffering and directio I was able to throttle the iops to a device to 10 per second using echo 8:32 10 > /sys/fs/cgroup/blkio/user.slice/blkio.throttle.{read|write}_iops_device


2

Since nginx 1.3.15 there is a "limit_req_status" directive that allows you to specify the http response code that the throttler will return. # Define a limit request zone called "bots" that will track requests by IP. limit_req_zone $binary_remote_addr zone=bots:20m rate=15r/s; # 429 = Too Many Requests limit_req_status 429; http status ...


2

bwlimit can be used only as a command-line parameter but not in /etc/rsyncd.conf


2

fail2ban allows custom filters and custom actions. You can create a custom filter to parse appropriate logs for successful ssh login with appropriate action. You can find many examples of different filters in /etc/fail2ban/filter.d/ and many examples of different actions in /etc/fail2ban/action.d/ If you don't want to use IP blocking then you can lock a user ...


1

Is throttling server performance ever used by companies to conserve power? Actually that is done all the time: all modern systems use power management und dynamic clock control to conserve power and reduce cooling. However, this is a dynamic process, controlled by processing demand. As soon as there's processing demand the clock ramps up, so you'll barely ...


1

I have found this thread on the exim mailing list. They discuss different hacks and workarounds for throtteling including writing a shell script to manually sleep and deliver the messages (with exim -M) postfix as a "smarthost" which apearently can do what we want configuring a router that just sleeps for some seconds and then passes the message to the ...


1

Context: ICMP is not a stateful protocol. In a scenario where you are sitting behind a router that NATs your traffic and you are pinging a device at the other side of the router, the router needs to keep track of your ICMP echo requests in order to let the respective ICMP echo replies in. How to do this if ICMP is not stateful? Just write an entry in the ...


1

There are recommendations in RFC 5508 "NAT Behavioral Requirements for ICMP", as to how long a router should retain state for ICMP packets, but there's no guarantee that anyone follows them.


1

I got the same result as you, but found that if I used blkio.throttle.{read|write}_bps_device that will limit throughput when the device is being used as a filesystem.


1

If you've looked into tc, then you're probably right, still, best to indicate the contents of running: tc qdisc show tc filter show dev eth0 (Adjust with your appropriate device, pp0 or whatever) Also, take a look at iptables with: iptables -nvL iptables -t nat -nvL If you don't find any tc or iptables rules responsible, then it could just be a problem ...


1

As far as I know AJP neither on the Apache nor on the Tomcat end does throttling, but instead tries to pass the request on between the layers as fast as it can. Here are some thoughts: One can set connection limits for workers, which when reached leads to the worker not accepting more connections, so as to avoid saturation and overload. One can introduce ...


Only top voted, non community-wiki answers of a minimum length are eligible