[GitHub] [incubator-pagespeed-mod] Ungov77 commented on issue #2089: High TTFB with PageSpeed

[GitBox <gi...@apache.org>]

Ungov77 commented on issue #2089:
URL: https://github.com/apache/incubator-pagespeed-mod/issues/2089#issuecomment-1095809062

   Hi, thanks for the reply
   
   The server is a dedicated server in OVH Datacenter in Beauharnois, Quebec, Canada
   Intel(R) Xeon(R) CPU D-1541 @ 2.10GHz (16 core(s)) - 2499 MHz
   64 GB Ram
   NVME SSD
   1GBPS network
   
   CPU load is always under 25% https://prnt.sc/Xte_RHbEGSgR
   
   We also integrated a benchmark into the website to calculate the page load speed of the different PHP modules and calculate SQL queries execution time. You can see it on the bottom of the page when adding ?benchmark=1 to the URL
   https://www.keinegotterkeinemeister.de/tshirt-sabotage-black-cat-D01004174288P0210/?benchmark=1
   
   Currently, it takes 0.2110 seconds to generate the page with PHP on the first page load. But out of this 0.21 seconds, 0.1865 is due to calling an external geolocation API. This is just because Cloudflare is currently disabled for our testing phase. When Cloudflare is enabled, we get the geolocation from $_SERVER["HTTP_CF_IPCOUNTRY"] so we avoid calling the external API and page load time goes down to around 0.01s
   
   TTFB is not the only problem, we also have high LCP and it throws warnings in Google Search Console: https://prnt.sc/2PjB74ekUgdR
   Our search ranking dropped a lot and we lost 50% of our search traffic since June 2021 and we suspect it could be partly due to that
   
   How can I confirm whether the disk is I/O bound or not? I followed Plesk guide and here are the results:
   https://support.plesk.com/hc/en-us/articles/360006437794-Plesk-for-Linux-server-is-running-slow-too-much-IO-interactivity
   `[root@ns508279 ~]# dd if=/dev/zero of=/root/testfile bs=512 count=5000 oflag=dir                                                                              ect
   5000+0 records in
   5000+0 records out
   2560000 bytes (2.6 MB, 2.4 MiB) copied, 13.634 s, 188 kB/s
   [root@ns508279 ~]# dd if=/dev/zero of=/root/testfile bs=512 count=5000 oflag=direct
   5000+0 records in
   5000+0 records out
   2560000 bytes (2.6 MB, 2.4 MiB) copied, 14.1703 s, 181 kB/s
   [root@ns508279 ~]# dd if=/dev/zero of=/root/testfile bs=512 count=5000 oflag=direct
   5000+0 records in
   5000+0 records out
   2560000 bytes (2.6 MB, 2.4 MiB) copied, 14.0026 s, 183 kB/s
   `
   This seems pretty slow. Could it be because we run pagespeed from too many domains? 
   
   Latency seems fine:
   ```
   [root@ns508279 ~]# dd if=/dev/zero of=/root/testfile bs=1024M count=1 oflag=direct
   1+0 records in
   1+0 records out
   1073741824 bytes (1.1 GB, 1.0 GiB) copied, 2.05548 s, 522 MB/s
   ```
   
   however iotop looks perfectly fine. Monitored it for around 10 minutes and IO never went above 1%
   https://prnt.sc/F65mP1MkTeBC
   
   also no high %WA with top:
   https://prnt.sc/Oe1hhgcqrNDC


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: dev-unsubscribe@pagespeed.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org