Update notes for 2019-03-31

content/posts/2019-03.md

@@ -1019,5 +1019,19 @@ $ time http --print h 'https://cgspace.cgiar.org/rest/items?expand=metadata,bits
     - Regarding RAM, our JVM heap is 8GB and we leave the rest of the system's 32GB of RAM to PostgreSQL and Solr buffers
     - Seeing as we have 56GB of Solr data it might be better to have more RAM in order to keep more of it in memory
     - Also, I know that the Linode block storage is a major bottleneck for Solr indexing
+- Looking at the weird issue with shitloads of downloads on the [CTA item](https://cgspace.cgiar.org/handle/10568/100289) again
+- The item was added on 2019-03-13 and these three IPs have attempted to download the item's bitstream 43,000 times since it was added eighteen days ago:
+
+```
+# zcat --force /var/log/nginx/access.log /var/log/nginx/access.log.1 /var/log/nginx/access.log.{2..17}.gz | grep 'Spore-192-EN-web.pdf' | awk '{print $1}' | sort | uniq -c | sort -n | tail -n 5
+     42 196.43.180.134
+    621 185.247.144.227
+   8102 18.194.46.84
+  14927 18.196.196.108
+  20265 18.195.78.144
+```
+
+- I will send a mail to CTA to ask if they know these IPs
+- I wonder if the Cocoon errors we had earlier this month were inadvertently related to the CPU steal issue... I see very low occurrences of the "Can not load requested doc" error in the Cocoon logs the past few days
 
 <!-- vim: set sw=2 ts=2: -->