Heap fragmentation in a long running Ruby process



In a long-running ruby process with a highly dynamic object-space, we encountered performance degradation and finally memory-allocation failure due to heap fragmentation. The problem can be mitigated by linking ruby against ptmalloc3.


Hi all! I’m writing this mail in the hope that my experiences may point you in the right direction, if you ever encounter a similar problem. Naturally I would be delighted to read your comments and advice on my conclusions and the steps taken.


http://ch.oddb.org [1] provides information on the swiss health-care market. Behind an Apache/mod-ruby setup lies a single ruby-process, which acts as a DRb-Server. Predating Ruby on Rails, the application is based on self-baked libraries [2-4].


A couple of weeks ago we experienced a spike in user requests. Although the application seemed to scale well most of the time, we began experiencing outages after a couple of hours. Whenever that happened, CPU-Load rose to 100% and DRb-Requests were hanging, sometimes for several minutes. At the same time, memory usage started rising considerably. If left to run for enough time, the application would crash with a NoMemoryError: ‘Failed to allocate Memory’ – even though there was still plenty of Memory available in the system.


Thanks to Jamis Buck [5] and Mauricio Fernandez [6] I was able to determine that the application was stuck for several seconds in glibc’s realloc, which may be called (via ruby_xrealloc) from basically anywhere within ruby where a new or enlarged chunk of memory might be required.


Having stated the diagnosis: heap fragmentation [7], there were a couple of things I could try to improve the performance of our application, all revolving around the principle of creating fewer objects, and in particular fewer Strings, Arrays and Hashes. By eliminating a number of obvious suspects (mainly to do with the on-demand sorting of values stored in a large Hash), I was able to raise the life-expectancy of our application considerably – close, but no cigar.


And then – all praise bugzilla – I found a bugreport [8] describing almost exactly my problems and leading me to ptmalloc3 [9]. Glibc’s malloc implementation is based on ptmalloc2, and may be replaced by simply linking ruby against ptmalloc3.


As far as I understand, ptmalloc3 does not eliminate heap fragmentation. However, due to the bit-wise tree employed in the newer version, it finds free chunks of the right size in shorter time by several orders of magnitude. Additionally, it seems that glibc 2.5 abandons its attempts to find a best-fit chunk after a while (possibly after 10000 tries), instead expanding the heap as long as possible and finally failing to allocate memory – causing first the fast rise in memory usage and later the observed NoMemoryError.


At this time, http://ch.oddb.org has run – powered by ruby and ptmalloc3 – for a little more than 24 hours without displaying any of the signs I have come to associate with heap fragmentation. Significantly less time is spent in allocating memory – and consequently in GC, and the overall memory-footprint has decreased by about 30%.


I hope this is of use – thanks in advance for any thoughts you want to share.

Hannes Wyss

[1] Open Drug Database
[2] Object-Database Access and Object Cache
[3] State-Based Session Management
[4] Component-Based Html generator
[5] Inspecting a live ruby process, Jamis Buck
[6] Ruby live process introspection, Mauricio Fernandez
[7] Heap fragmentation, Bruno R. Preiss
[8] Glibc bugzilla report 4349, Mingzhou Sun, Tomash Brechko
[9] Ptmalloc home, Wolfram Gloger

ODBA Verbesserungen

Aus dem Mail von Hannes Wyss an die ywesee interne Liste:

Der Aktuelle commit der ODBA
beinhaltet mehrere Verbesserungen und zwei Bugfixes, die insgesamt die
langfristige Memory-Auslastung kontrollieren sollen. Endgültige
Bestätigung werden wir erst aus dem online-Dauerbetrieb erhalten.

  • der Cleaner-Thread im ODBA.cache läuft mit höherer (normaler) Priorität und häufiger, dafür aber für kürzere Zeit. D.h. konkret dass im Zeitraum von ca 10 Sekunden jeweils 500 Objekte überprüft und gegebenenfalls aus dem Cache gelöscht werden.
  • Bugfix: wenn Collection-Elemente einzeln aus der DB geladen werden, werden sie neu auch im Cache registriert.
  • CacheEntry führt darüber Buch, welche Objekte auf ein bestimmtes anderes Objekt zugreifen. Neu wird dies nicht mehr direkt über die Referenz gemacht, sondern über odba_id/object_id – damit ist diese Information garantiert kein Hindernis für die GC mehr – bis jetzt wurden die Referenzen jeweils rechtzeitig entfernt und ‘sollten’ auch keine Rolle gespielt haben, jetzt gibts dafür eine Garantie.
  • Ebenfalls aus CacheEntry entfernt wurde der @collection-Eintrag. Beim Speichern einer Collection ist es notwendig zu wissen, welche Elemente der Collection bereits in der Datenbank liegen, welche gelöscht werden müssen, und welche neu hinzukommen. Dies wurde bis jetzt mit eben diesem @collection-Eintrag gelöst. Neu werden die bestehenden Daten jeweils direkt von der DB bezogen.
  • Bugfix: die Ausgabe einer Fehlermeldung führte bei einer speziellen Konstellation zu einem Memory-Spike. (konkret: Narcotic#to_s ist abhängig von den Substanzen in Narcotic@substances. Bei einigen instanzen von Narcotic war das entsprechende Objekt aber gelöscht oder nie gespeichert worden. Dies sollte eigentlich mit einer Fehlermeldung vermerkt werden; da die Fehlermeldung aber ODBA::Stub@container.to_s als Bestandteil hatte, ergab sich ein unendlicher Loop von Exceptions) Dieser Bug nahm die Hälfte der 3 Tage in Anspruch. Ich konnte ihn schlussendlich nur dank dieses Tools finden.

More on Ruby Memory-Leak

I found some more interesting posts:

  1. Memory leaks in my site
  2. Rails memory usage case study
  3. bleak_house
  4. Finding open file descriptors: ‘lsof’ is a neat utility for listing open ports, sockets and files on a per process basis.
  5. The WeakRef and WeakHash.

Still I put it as following: I believe Ruby has a serious memory leak problem that is not been taken seriously. It is not even only about the memory leak. It is the GC that takes to long once the memory reaches a certain seize. Once our Application uses more then 2 GB of Memory the GC takes more then 50 secs to do its job. Then we have to do a restart of our Application because our Application should not be unresponsive for the User for 50 secs or more. PS: We got 12 GB of memory for Ruby on our server.