From: Dave Sill (MaxFreedom_at_sws5.ornl.gov)
Date: Wed, 24 Aug 2005 16:03:37 -0400
I've got an RHEL 3 server running a vsftpd FTP daemon. We're seeing
occasional temporary hangs when certain systems retrieve certain
files. For example, from a particular Solaris client:
---- root@client# ftp server Connected to server. 220 NOTE: Please login as user ftp Name (server:root): ftp 331 Please specify the password. Password: 230 Login successful. ftp> bin 200 Switching to Binary mode. ftp> prom Interactive mode off. ftp> cd test1 250 Directory successfully changed. ftp> mget * 200 PORT command successful. Consider using PASV. 150 Opening BINARY mode data connection for sgpmfrsrE3.b1.20040602.000000.cdf.v0 (816156 bytes). 226 File send OK. local: sgpmfrsrE3.b1.20040602.000000.cdf.v0 remote: sgpmfrsrE3.b1.20040602.000000.cdf.v0 816156 bytes received in 0.48 seconds (1665.80 Kbytes/s) 200 PORT command successful. Consider using PASV. <delay of several minutes here> 150 Opening BINARY mode data connection for sgpmfrsrE3.b1.20040602.180300.cdf.v0 (297204 bytes). 226 File send OK. local: sgpmfrsrE3.b1.20040602.180300.cdf.v0 remote: sgpmfrsrE3.b1.20040602.180300.cdf.v0 297204 bytes received in 93 seconds (3.14 Kbytes/s) ftp> by 221 Goodbye. root@client# ---- Once a session hits the delay point indicated above, further attempts to ftp from the client to the server hang: root@client# ftp server Connected to server. <delay of here, sometimes permanent> Attempts to connect from other clients during this lull experience no delays. Things we've discovered: - grabbing the files individually with get in the same order produces the same hang - grabbing the files in the reverse order doesn't hang - renaming the files has no effect - replacing the file contents with /dev/random or /dev/zero contents of the same size has no effect - retrieving the same files from another client running the same OS release on the same hardware doesn't hang - serving the same files from another server running the same OS on the same hardware doesn't hang - files that some offsite user is having trouble retrieving don't exhibit the same problems internally, even from the known problematic client - the problem is highly repeatable for a given client system and set of files, but rarely even these cases proceed normally Has anyone ever seen anything like this? Any ideas what could be causing it or how to debug it? -- Dave Sill Oak Ridge National Lab, Workstation Support Author, The qmail Handbook <http://web.infoave.net/~dsill> <http://lifewithqmail.org/>: Almost everything you always wanted to know.