Sometimes a Ping fails (ICS v5)

[Techo 0]

Hi folks! Thanks in advance for any tips/suggestions.

I'm simply using the TPing component for a while now. Works great overall. Trying to figure out why a ping might fail. Ever since upgrading to win7/8 , the ping happens to fail now and then, when it would have no reason to.  Using Size=56 , Timeout=200. TTL=64 and v5 of TPing.  WinXP to WinXP is fine 100% of the time. WinXP->Win7/8 or Win7/8->Win7/8 has a bit of failure rate.

 

Edited January 18, 2020 by Techo

[FPiette 383]

When it fails, try using command line ping to see if it fails as well AT THE SAME MOMENT.

 

[Clément 148]

Hi,

 

You can test with a tool I wrote. The community edition is free: http://www.dhs.com.br/dhsPinger.html

 

[Techo 0]

Thank you both for the replies. Much appreciated. Some good food for thought.

 

It's quite random. e.g. 1000th pings work fine and then fails on the 1001st, 1002nd is ok. Perhaps the frequent ping is causing some network traffic or back up. I will try the command prompt Ping , set it to 1200 count and look through the log.

 

Having XP->XP work perhaps says something? The newer OSes definitely have a slightly slower network refresh. I turned off all firewall and antivirus apps on some test machines.  As an alternative, I can also try changing the ping timeout to 500 and lower the frequency/rate of the ping. This is part of a utility that checks which computers are online before proceeding with the rest of the work. As a last resort I can map the target PC's on the network as drive letters and see if that helps produce a quicker connect. I did read somewhere that with other ping methods one must run the app under administrative rights to be able to ping out , so some layer of security are adding random delays. Will test with dhsPinger and write back.  Thanks again!

 

Edited to add: Will check to see if changing PCs running Win7/8 under IPv6 to IPv4 will help.

 

Edited January 19, 2020 by Techo

[FPiette 383]

You two involved PC are on a LAN? Same segment or is there routers or switches? Ping is not a reliable protocol, one ICMP message can be lost in case of high trafic. If you loose one ICMP message per one thousand, it looks acceptable to me.

 

 

[Techo 0]

1 hour ago, FPiette said:

You two involved PC are on a LAN? Same segment or is there routers or switches? Ping is not a reliable protocol, one ICMP message can be lost in case of high trafic. If you loose one ICMP message per one thousand, it looks acceptable to me.

 

 

I see! I wouldn't have thought this would be an acceptable rate.  With XP, the failure rate is almost non existent. I'll re-think my approach a bit. But Yes to the LAN and it's on one network switch/router segment.  I check for a ping back and then check for a fileexists flag before proceeding with the rest, as sometimes a ping works but the network isn't available for file access. That could be due to a corrupt Windows OS or bad hardware. The ping is just the first line of the test. But when it fails it requires a human operator to click the "restart scan" (so to speak). The failed ping is hard to catch, so perhaps a re-try ping is enough. Up to 3 auto re-tries should be enough? Thank you, Mr. Piette. It means a lot to get direct feedback from you. I'll research this some more.

Edited January 19, 2020 by Techo

[FPiette 383]

Before stating remote is down, you'd better do a few pings and decide the remote site is down only if all pings are failing or at least a given percentage of the pings failed.

Please be aware that it is prossible to completely shutdown ping response in a system. If you control both servers then you can use ping. If you don't, be preapared to use an alternative the day ping are disabled.

 

[Fr0sT.Brutal 900]

3 hours ago, FPiette said:

Before stating remote is down, you'd better do a few pings and decide the remote site is down only if all pings are failing or at least a given percentage of the pings failed.

Please be aware that it is prossible to completely shutdown ping response in a system. If you control both servers then you can use ping. If you don't, be preapared to use an alternative the day ping are disabled.

 

+1, for this reason, it's more reliable to check for a concrete service rather than just online status of a host. It could be online but a needed service could be down or pings could be disabled while host is up and service is running.

[Techo 0]

Yes, I can configure both PC's. I also noticed I left the default TTL to 64. I'm reading that Windows OSes would probably prefer the default start at 128. Thanks again! I'll write back soon. Adding on: This is an internal network so perhaps the TTL set to a default 64 most likely isn't the culprit but it's something to consider. 

Edited January 20, 2020 by Techo

[Guest]

FPiette is right under heavy traffic the ICMP is first to go then udp then TCP but tcp packet will be resend because they can !

Windows 7 for me was always worse than Windows XP in connection performance and speed, but if you tried to remove/disable the IPv6 from Windows 7 then it will measure up to Windows XP

If TTL will affect the ICMP delivery in NAT then you have real problem with your Windows drivers and may be other user mode software installed on it, check that,.. as ICMP on NAT should not require more than 5 ms or the router is dropping them because it is busy or low on memory, and by busy i mean %50 of its CPU, if your router have web interface to report status, then check that also, at %50 expect more than %70 drop of UDP and forget about ICMP completely.

 

[Techo 0]

Quick update. Changing the TTL to 128 didn't help. Will check if using IPV4 changes anything, even though I'd prefer not to but for the sake of trouble shooting...

[Techo 0]

Another quick update. Will test on peer to peer Win10 pc's. And see if that helps. Cheers!